About
I am a builder who has been leading technical teams for many years. I dabble in a lot of…
Activity
2K followers
-
Ryan Parman posted thisAs I integrate more AI into my software development practices, I've been going slowly, making sure I understand each step, performing manual code reviews, and working to ensure that my engineering skills grow instead of atrophy as I adopt AI. 1. I strongly prefer Kiro IDE over other tools I've used specifically for its built-in spec-driven development flow. Requirements → technical design → tasks to perform. Human *driving* the loop throughout. I'll use Claude Code or OpenAI Codex if I'm just exploring an idea or want a proof-of-concept, but Kiro when I want to do real work. 2. Any code that is written is automatically run through a gamut of linters, static analysis tooling, security checks, etc., so that the agent can self-correct as it goes. I also have a fairly specific set of instructions per language (e.g., Bash, Python, Go) as well as use cases (e.g., building a CLI vs. building a server) and favored libraries (pelletier/go-toml over BurntSushi/toml). If the AI makes a mistake, I update its steering file. 3. All functions and complex sections of code have code comments focusing more on WHY than WHAT. I always review these for correctness before committing. 4. I use KiroGraph <https://lnkd.in/gKBjcuGR> as a per-project knowledge graph, which can dramatically speed-up code knowledge for agents, preventing the need to re-assess the codebase from scratch every time. It learns about your project and assists the efficiency of the agent, and significantly reduces token usage. 5. I generate (and then manually correct upon review, if necessary) a QuickStart doc, as well as a Comprehensive Review doc, which live in the `docs/` folder in the repository. Both are incredibly useful for both agentic and human collaborators. 6. I never, ever permit the agent to commit anything. That is a human-only action. No exceptions. 7. Now, I'm building an "adversarial citation gate," which is a fancy name for fact-checking an agent in real-time, preventing it from hallucinating. During Kiro's requirements and design phase, if the topic is important/critical (e.g., auth, security, data integrity), the fact-checker will kick-in with a non-LLM step and fact-check the agent's claims against the provided spec docs — escalating to human judgement if they begin to argue with each other. It's presently Kiro-only and EXPERIMENTAL. I'm currently in the depths of putting it through the paces. Check it out if you're interested. I call it "Magneto". Phases 0, 1, and 2 have been implemented and are undergoing hands-on usage testing. * https://lnkd.in/g-mQZJix * https://lnkd.in/ggHNs9YP
-
Ryan Parman reposted thisRyan Parman reposted thisAfter dodging multiple RiFs over nine years, I finally got got. If you know of anyone looking for a Technical Program Manager with experience working in the mobile, firmware, and cloud spaces, let me know.
-
Ryan Parman shared thisWith GCP Terraformer having shutdown in recent months, is anybody aware of alternate tooling for generating Terraform code from real infrastructure? https://lnkd.in/gRXvJvXa
-
Ryan Parman reposted thisRyan Parman reposted this💙 My journey with Bungie has ended today along with hundreds of my peers. Working on the storied Destiny franchise has been a privilege and my colleagues are the absolute best in the industry. I am excited to consider new opportunities in community management, social media strategy, and influencer marketing! 🤝 Connections: Your boosts of my past hiring posts generated 1000+ referrals of those affected by layoffs - this time it's me asking for your support. Please like, comment, and share this post to help me increase its reach on LinkedIn with #HireEvanBerman. 👋 Recruiters: I'm open to remote or hybrid (greater San Diego area) positions. Let's chat! 💠 To our community: thank you, you know, for being my Guardians. #opentowork #hiring #layoff #layoffs #bungie #communitymanager #communitymanagement #socialmediamarketing #socialmedia #communitydirector #influencer #influencermarketing #contentstrategy #communitydevelopment #community #gamejobs #gameindustry #gamedev
-
Ryan Parman reacted on thisRyan Parman reacted on thisRejection post three. I'll keep posting my job rejections here not out of vindication or perceived pettiness, but rather as a log of the types of rejection. Rejected for a 9th time now. The employer did not specify why I was not chosen to interview, and then decided at that point it was okay to ask me to consider donating money to their organization in what I can only assume was an automated form letter.
-
Ryan Parman reacted on thisRyan Parman reacted on thisGot rejected at a job, saw the company post the same job again, so I applied. I decide when we're done.
-
Ryan Parman reacted on thisRyan Parman reacted on thisHere's a new one that gives me some hope for the interviewing process in tech. After having a final round interview last week, I got the email no one wants to receive -- a form rejection letter: "Thank you for taking the time to interview with us.... At this time, you were not selected to move forward." But this one something different I hadn't seen before. A sentence at the end that stated "Please feel free to call me at ###-###-#### if you would like to discuss the hiring decision" and it was signed by the recruiter. After taking some time to process my disappointment, I called the recruiter. She answered her phone and candidly read me the summary of the hiring huddle, both pros and cons. And the feedback felt fair -- they saw the areas where I felt I was strong and the concern that wound up being the deal breaker was the one interview where I didn't have concrete examples. The phone call took SIX minutes, probably 1.5 minutes of which was me trying to understand if it was worthwhile to apply for future roles given that gap. So basically FIVE MINUTES to treat a candidate with a little bit of respect and courtesy, and it left me feeling much better about the process, the company, and honestly more confident in myself despite the rejection. I know people are overwhelmed, but 10 phone calls like that would take an hour a week of someone's time. And it would probably save time in the long run as candidates are better prepared going forward.
-
Ryan Parman reacted on thisRyan Parman reacted on thisNobody tells you this part when you're coming up in tech. Being a Senior/Staff engineer isn't about being deep in one thing. It's about being able to go deep in anything, on demand, at any time. Here's an honest rundown of what crossed my desk this week: AWS: Closing out old accounts in Organizations, S3 replication rules and bucket policies, decommissioning an Aurora cluster (snapshot first, always), tearing down a Redis ElastiCache cluster, chasing down ALB target group health, managing ACM certs, updating IAM OIDC role policies, fixing CodeArtifact permissions for Maven publishing, digging through SSO permission sets, and general EC2 triage. Azure: RBAC role assignments, Contributor vs Reader, the usual "who can touch what" conversation. GCP: Standing up a new project inside our org folder structure, building out a Pub/Sub topic and push subscription, writing a least-privilege IAM binding for it. Then the stuff that doesn't fit neatly in a cloud console: A WordPress multisite redirect loop that took real diagnosis to run down, Apache vhost configs and .htaccess rewrites, an SSL termination fix buried in wp-config.php. And the glue that holds it all together: Writing a new Terraform module and getting it through review, managing PRs on GitHub, reviewing a Bitbucket pipeline for OIDC deployment, automating ticket updates through the Jira REST API, and tracing DNS resolution when something just wouldn't propagate. Three clouds. An app stack diagnosis. CI/CD tooling. All in one week. This is what "senior" actually looks like day to day. Not a title. Not a pay grade. It's the ability to context-switch across an entire stack without losing the thread, because production doesn't care what you specialize in. If you're early in your career and feel behind because you don't know everything...nobody does. You build this range one incident, one ticket, one "why is this broken" at a time. #SRE #DevOps #CloudEngineering #AWS #Azure #GCP #Terraform #PlatformEngineering
Experience
Education
-
Carrington College
-
-
Activities and Societies: N/A
Attended when the school was called "Silicon Valley College" in San Jose, CA. It offered a Bachelor of Arts degree in "Design and Visualization," which was discontinued in January 2013.
Licenses & Certifications
-
Deep Dive: Building Your First Application
Palantir Technologies
-
Deep Dive: Creating Your First Ontology
Palantir Technologies
-
Foundry & AIP Builder Foundations
Palantir Technologies
Patents
-
System and methods for user authentication across multiple domains
Filed US US20160241536A1
A new approach is proposed that contemplates systems and methods to support verification of a user's authentication information across multiple websites/domains owned and/or operated by different entities, which share users during a single session. When the user attempts to login to a first website/domain, he/she is required to provide authentication information in addition to user-id/password. An authentication platform is configured to generate and communicate the additional authentication…
A new approach is proposed that contemplates systems and methods to support verification of a user's authentication information across multiple websites/domains owned and/or operated by different entities, which share users during a single session. When the user attempts to login to a first website/domain, he/she is required to provide authentication information in addition to user-id/password. An authentication platform is configured to generate and communicate the additional authentication information to the user and verify the additional authentication information the user provided to the first website/domain. When the user later attempts to access a second/unrelated website/domain, the verified additional authentication information is provided by the first website/domain to the second website/domain in the form of a signed cookie. The second website/domain parses the cookie and provides the additional authentication information to the authentication platform for verification without requiring the user to input it again at the second website/domain.
Other inventorsSee patent -
Hive-based peer-to-peer network
Issued US WO/2008/033424
A collaborative data transferring process can combine segments from all known servers and peer-to-peer (P2P) sources simultaneously, regardless of their native protocols. The process uses variable data block size that can be dynamically selected according to sizes provided by sources, e.g., according to the protocol of the source, and can generate hash values or validation codes on the fly so that compliance with validation techniques (if any) of other protocols is not required. The process may…
A collaborative data transferring process can combine segments from all known servers and peer-to-peer (P2P) sources simultaneously, regardless of their native protocols. The process uses variable data block size that can be dynamically selected according to sizes provided by sources, e.g., according to the protocol of the source, and can generate hash values or validation codes on the fly so that compliance with validation techniques (if any) of other protocols is not required. The process may be classified as a P2P protocol, although it also contains centralized elements. Machine language implementations and low syntax overhead allow file exchanges over a homogeneous network with high throughput and low bandwidth consumption.
Other inventorsSee patent
Projects
-
AWS SSO Manager
- Present
Streamline managing AWS Identity Center (née AWS SSO) profiles.
When you work with dozens, hundreds, or even thousands of AWS accounts, staying on top of everything can be exceptionally challenging. As accounts get added, removed, or renamed in AWS organizations, you need a way to stay on-top of the accounts, permissions/roles, account IDs, and more. AWS SSO Manager helps streamline this process, works alongside (but does not require) AWS CLI and tools like AWS Vault. -
Linux Packaging
Working on standing up a pipeline for producing Linux packages for a variety of software for Alpine Linux, Debian/Ubuntu, and Fedora/Amazon Linux with both x86_64 and arm64 builds. I use many bits of software that are not available as Linux packages, so this is my attempt at creating it, and automating the building of updates.
-
Corefunc (Terraform/OpenTofu provider)
Corefunc is a set of functionality that I wish were built into Terraform/OpenTofu. Corefunc provides a series of functions and makes them available as Terraform Data Sources, Terraform Provider-Defined Functions, and as a Go library that can be imported in your Terraform integration tests (e.g., via Terratest). This not only enables functionality that Terraform/OpenTofu are missing, but also enables identical functionality between your Terraform code and your Terraform test suite.
-
AWS “Hub and Spoke” CLI
One hub, many spokes, much security.
Simplifies the process of executing commands from the AWS CLI, Terraform, etc., on a spoke account, connecting through a hub account, in an AWS Organization. Most useful for teams with many AWS accounts configured as part of an AWS Organization, generally using AWS Identity Center (née AWS SSO).
No direct IAM credentials are required as an AWS Identity Center user assigned to an appropriate base role can assume a hub role, then assume a spoke…One hub, many spokes, much security.
Simplifies the process of executing commands from the AWS CLI, Terraform, etc., on a spoke account, connecting through a hub account, in an AWS Organization. Most useful for teams with many AWS accounts configured as part of an AWS Organization, generally using AWS Identity Center (née AWS SSO).
No direct IAM credentials are required as an AWS Identity Center user assigned to an appropriate base role can assume a hub role, then assume a spoke role, then execute software against AWS services adhering to the policy associated with the spoke role.
Languages
-
English
Native or bilingual proficiency
Recommendations received
34 people have recommended Ryan
Join now to viewExplore more posts
-
Piotr Retmańczyk
Everybody gangsta until the reality hit too hard. Me - as an engineer I need to be up to date with the trends and techs. I need to be useful, flexible. I have to understand the architecture, patterns, new tools in the whole SDLC and provide value each day! And all of that is perfectly fine! - this is part of the job. Part of engineer path. But! Why in the hell - and this is constantly repeating through the years over and over again and I can see it clearly - the business processes are completely not fitting well, the decisions are foggy, the plans change so rapidly that you feel like garbage and what is more tragic - in plenty of cases there are tons of undecisive leaders who need to fight the procedures too much (tech leaders / team leaders) and managers who are even more undecisive. Who can not take responsibility. Who are not assertive enough to say that something is absolutely wrong. I have a feeling that it is not about technology, product, AI or tech anymore (or is.. in SoMe). This is all about the unnecessary politics, sick processes, greed and leadership crisis... #leadership #crisis #responsibility #ethics #human #first
-
Rishav Goyal
Dhurandhar 2 isn’t a sequel. It’s a system upgrade. ~4-hour runtime. 2 AM shows. ₹35–40Cr previews. 8L+ tickets before release. That’s not hype. That’s trust compounding. Part 1 built it. Part 2 is cashing it. Bollywood is shifting: from films → franchises from stories → systems Back-to-back shoots. 5-language rollout. Continuity as retention. Same playbook as tech. They didn’t fit theatres. Theatres adapted to them. That’s power. Dhurandhar 2 isn’t chasing records. It’s rewriting the rules.
1 Comment -
George Cretu
A lot of engineers I speak with feel the same thing right now. They’re not underqualified. Not because they aren’t doing great work. But because the landscape keeps shifting faster than anyone can prepare for. One week a company wants one thing, the next week the role looks completely different. Expectations rise, clarity drops, and the whole process feels harder to trust. If you’re feeling that, you’re not alone. The best thing you can do is focus on what you can actually control: Be clear about what you want. Be honest about what you’re great at, and pay attention to how a company communicates during the interview, that usually tells you everything about what it’s like to work there. You don’t need to fit every possible box, you just need to find the place where your skills are respected and the expectations make sense, those roles still exist.
1 Comment -
Gunaranjan Sundararajan
I have 16 years in the trenches of IT, moving from a single dev to leading teams of 20+, and if there’s one thing I have learned, it’s that "Data Lakes" usually just turn into expensive swamps without a literal gatekeeper. if you aren't enforcing Data Contracts at the producer level, you’re basically just waiting for a Friday afternoon to turn into an all-night outage lol we have all been there right ! Some dev in a different team changes some random field for their sprint, Suddenly, your analytics are bleeding red. If your strategy is clean it up in the warehouse, you aren't an architect ! Validation belongs at the boundary of the microservice. If the event doesn't match the Avro schema in the registry, it gets dead-lettered. Period. You can't have "independent teams" if every minor schema change breaks other departments. The contract is the only thing that makes autonomy possible. Don't have PDFs about PII and GDPR in shared drive alone , Embed those rules (masking, encryption, SLAs) directly into the YAML spec too. In a real Data Mesh, the Data Guard is the only thing standing between a scalable system and total architectural chaos. Where there's a will, there's a way ! Hope ya'all Have a great week ahead ! #systemdesign #datamesh #dataengineering
-
Gabriel Radu
A slightly philosophical, but really important consideration for scaling agentic development today... Lately I've been thinking a lot about who should own "the outer harness", i.e. the collection of skills, agents, evals and workflows you build around an "inner harness" such as Claude Code, Codex or Cursor. Just weeks ago I would have said that in a large enterprise this becomes another platform team, and that you should absolutely invest in such a team if you are serious about adopting agentic development at scale. But after talking to both colleagues and clients, I'm not so sure. Yes, we need principles and some sort of reference architecture for what a good harness looks like, but if we go full blown and fully centralise it, we risk the new team becoming a bottleneck. Agentic practices change very fast, and development flows are still highly adapted to each product or team, no matter how much we want to standardise. So it depends... You might have a lot of small, non-business-critical applications where you genuinely just need high feature throughput, and which may be suitable for adopting a centralised harness. But you may also have large business-critical applications where teams will need to build and maintain their own. If you think about it, it's simply applying the Minimum Viable Platform concept from Team Topologies, and how much and what to centralise is something every CTO needs to spend serious time considering.
-
John Gallagher
Treating observability as an investment is the smart mindset. Instead of focusing on the price tag alone, I ask how much ROI I'm getting on that data. Personally, and I know this is hard to measure, but the missing gap right now is having tools to measure the effectiveness of all that data you're collecting. As a low bar, is it ever queried? Is it ever searched on? Is that data ever acted on? Is it used? Because data that's not been queried on for months if not years isn't contributing positively to your return on investment. As usual, Charity Majors speaks the true true.
1 Comment
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top contentOthers named Ryan Parman in United States
5 others named Ryan Parman in United States are on LinkedIn
See others named Ryan Parman