Aerospace × AI/ML Engineer · Hyderabad
Fresher ML Engineer with B.Tech Aerospace Engineering (VIT Bhopal) + PG Certificate AI & ML (IIIT Hyderabad). Bridging domain-deep aerospace knowledge with LLMs, RAG pipelines, PINN physics modelling, and orbital mechanics — end to end.
Live Platform
Full-stack aerospace intelligence platform — FastAPI backend, Next.js Mission Control dashboard, orbital Digital Twin, conjunction engine, Knowledge Graph, GraphRAG, and multi-agent AI. 557/557 tests passing. Deployed on Railway + Vercel.
Production-grade Space Situational Awareness platform. 50K+ object orbital globe with GPU instancing, live conjunction detection, Digital Twin propagation, Knowledge Graph with vis-network, GraphRAG with Anthropic, multi-agent AI workflows, JWT RBAC auth, Prometheus + Grafana observability. Built across 15 engineering phases.
Flagship Demos
Three client-side aerospace visualization demos — single-file, no backend, publicly deployed — covering space situational awareness, orbital intelligence, and live threat monitoring. (For the full-stack production platform, see ORBITIQ-X below.)
Orbital intelligence command centre with live density heatmaps, a Knowledge Graph canvas, nine SSA modules, and an AI Copilot — backed by a Cloudflare Worker proxy for real-time space weather and conjunction data.
Single-file 140KB HTML. SGP4 propagation, ISS/Starlink/NavIC real-time tracking, 3D globe, pass predictor (Hyderabad default), asteroid intelligence via NASA NeoWS, space weather overlays.
Eight-module live ops platform. Real-time ISS tracking, Chart.js visualisations, electric amber on ops-navy branding — monitoring threats and trajectories in continuous ops mode.
Projects
Enterprise operating platform for MMDI. Next.js 16, React 19, TypeScript, Supabase — 31 workspace modules across Operations, Manufacturing, Finance, and Compliance, plus an AI Copilot with 16 grounded tools and a from-scratch cost-estimation engine.
LoRA fine-tuned Salesforce/CodeGen-350M on A100 GPU. Four tasks: program synthesis, documentation generation, commit messages, Python↔Java/C++ translation. +34pp CodeBLEU across tasks.
Physics-Informed Neural Network for aeroacoustic drone detection. 2D wave equation governing model, Fourier-feature PINN, FFT/STFT signal processing, Streamlit dashboard — deployed to Streamlit Community Cloud.
India's open-source aerospace industry intelligence platform. Oracle DDL schema, .NET 8 Web API (60+ endpoints), Bootstrap 5 frontend, 10 modules — 87 Indian aerospace companies with CEOs, live projects, job openings.
600+ platform aerospace intelligence database — single-file HTML (~1MB). Full detail panels, Wikipedia REST API integration, offline AI fallback covering Skyroot, Tejas, BrahMos, AMCA, PSLV, DRDO UAVs, S-400.
Two-phase educational SPA covering 24 space data modules. All assets inlined as SVG (no external hotlinks). Deployed to GitHub Pages — covers orbital mechanics, SSA, GNSS, remote sensing and more.
Capabilities
Background
Built EKMS ("MMDI ONE"), a 31-module enterprise operating platform (Next.js 16, React 19, Supabase) for MMDI's operations, manufacturing, and compliance workflows, plus an AI Copilot with 16 grounded tools.
Led Group 39 capstone project. LoRA fine-tuned CodeGen-350M (r=16, α=32) on A100 GPU. +34pp CodeBLEU improvement (72% absolute) across four code generation tasks. Supervised by Prof. Anil Neelkanti; mentored by Pawan Baswani and Nayan Anand.
Acoustic detection system for counter-drone applications. Composite manufacturing (VARTM), NDT inspection, STM32 embedded integration. Contributed to signal processing pipeline for drone acoustic signature classification.
ML Engineer / GenAI / Aerospace AI roles. Based in Hyderabad — open to remote. Target domains: defense-adjacent autonomy, SSA, and aerospace intelligence systems.
Expected CTC · ₹7–10 LPA (Fresher)