██╗ ██████╗ ███╗ ███╗ █████╗ ███████╗██╗ ██╗ ██████╗ ███████╗██╗ ██╗ █████╗
██║ ██╔═══██╗████╗ ████║██╔══██╗██╔════╝██║ ██║ ██╔══██╗██╔════╝██║ ██║██╔══██╗
██║ ██║ ██║██╔████╔██║███████║███████╗███████║ ██████╔╝█████╗ ██║ ██║███████║
██║ ██║ ██║██║╚██╔╝██║██╔══██║╚════██║██╔══██║ ██╔══██╗██╔══╝ ██║ ██║██╔══██║
███████╗╚██████╔╝██║ ╚═╝ ██║██║ ██║███████║██║ ██║ ██║ ██║███████╗███████╗██║██║ ██║
╚══════╝ ╚═════╝ ╚═╝ ╚═╝╚═╝ ╚═╝╚══════╝╚═╝ ╚═╝ ╚═╝ ╚═╝╚══════╝╚══════╝╚═╝╚═╝ ╚═╝
⚡ SYSTEM LOG ────────────────────────────────────────────────
> booting lomash.exe ...
> loading modules: [ros2] [mujoco] [pytorch] [flutter] [fpga]
> affiliation: M.Tech Autonomous Systems & Machine Intelligence
> @ ABV-IIITM Gwalior (CGPA 9.78/10)
> prior training: B.Tech CE, CHARUSAT (CGPA 9.41/10)
> former: intern @ ISRO - Space Applications Centre
> status: RESEARCHER MODE - ENGAGED
> current_mission: VLAs and Multimodal Multilingual Langual Models
──────────────────────────────────────────────────────────────
AI Engineer working at the intersection of computer vision, LLMs, and robotics - building and deploying models that go from simulation to real hardware. Graduate Researcher (MTP student) at ABV-IIITM Gwalior, developing vision-based control pipelines for the SO-101 robotic arm and researching Vision-Language-Action (VLA) models for autonomous planetary manipulation.
Previously trained at the ISRO Space Applications Centre, deploying quantized deep learning models for real-time edge inference on planetary rovers.
class Lomash:
role: "AI Engineer - Computer Vision, LLMs & Robotics"
callsign: "lomash_irl"
focus: ["computer vision", "LLMs/VLMs/VLA", "edge AI deployment"]
hobby: ["low-cost hardware", "applied AI"]
status: "IEEE Space 2026 Ambassador""A Jedi's strength flows from the Force. Mine flows from a well-tuned loss curve."
- 🦾 Eye-in-Hand IBVS for SO-101 Arm (MuJoCo, OpenCV, Python) - Image-Based Visual Servoing with ArUco detection and Chaumette's pixel-based control, converging reliably to 30–40 px feature error, with joint/velocity limits, automatic divergence recovery, and a JSON-based teaching tool
- 🧭 SO-101 control systems + VLA research - mathematical control formulations in MuJoCo, plus Vision-Language-Action models for autonomous planetary manipulation, under the SERB-backed bimanual on-orbit testbed program
- 📸 Structure-from-Motion pipeline - 3D-scanning with the SO-101 gripper camera, Depth Anything 3, Open3D, damped least-squares IK
| Project | Description |
|---|---|
| 🤖 AI-Powered Planetary Rover (Jan–Jul 2025) | Raspberry Pi 4B rover on Bookworm OS, 65°C thermal cap, object detection + monocular depth estimation, dual autonomous/VNC-manual navigation |
| 🌪️ Tropical Cyclone Intensity Regression (IITM Pune, Aug–Nov 2024) | 4-channel deep learning pipeline mapping satellite imagery to cyclone intensity - published in the WGNE Blue Book |
| 🧠 Knowledge Extraction System using LLMs | RAG pipeline, FastAPI + LangChain, text-to-knowledge transformation |
| 💼 Full Stack @ Capitalize (June 2024) | Flutter + FastAPI apps, Selenium data pipelines with rotating proxies at 99% success rate |
- 🎖️ IEEE Space 2026 Ambassador - representing IEEE space initiatives; also Student Branch Chapter Secretary, IEEE MTTS IIIT Gwalior
- 👨🏫 Teaching Assistant, ABV-IIITM Gwalior
- 📈 Academic Rankings - KEA PGCET 2025: Rank 52 · GATE 2025 (Data Science & AI): AIR 4687
- 📚 Publications - IEEE Xplore (2025, 2026) · WGNE Blue Book (2024) · Springer research collection (2024)
- 🎤 Conference Presentations - IEEE SPACE 2026 · InGARSS 2025 · PRL Venus SC 2024 · ICHCSC 2024
"May the gradients converge and the loss curves hold." 🤖⚡