👋 Hello, I'm Shubhro Dev an AI/ML researcher working on large language models, multimodal deep learning, natural language processing, agents and time-series analysis.
🎓 B.Tech in Computer Science and Engineering from RGIPT (an Institute of National Importance, along the lines of the IITs), 2022–2026
🔬 Most recently a Technical Research Intern with the Global DAI R (SDE-IN) team at Siemens, researching the validation and verification of LLM outputs
📝 I like building projects which have real-life impact or which sounds fun to me!!
🌐 Everything in one place: spearb0lt.github.io
🛠️ Selected Projects
| Project | What it is |
|---|---|
| Privacy-Preserving Predictive Maintenance | Federated, differentially private RUL prediction on the IDA 2024 SCANIA-X challenge. Spectral-DP and DP-SGD implemented from scratch with a Rényi/Moments accountant, a Transformer time-series encoder paired with a TabTransformer head, aggregated across heterogeneous clients with flwr. Best RUL model on the dataset at MSE 2725, while an advanced membership inference attack gets only 49.12% AUC — i.e. no better than a coin flip. |
| UHCI / SignBridge | A fully offline communication app for Deaf and non-speaking users. Twelve input channels — morse keyed by a blink or a tap, sign language, lip reading, eye gaze, air-writing, a head-driven cursor, braille held up to the camera — all feed one message box. No API keys, no accounts, no network at runtime. ~129 modules, 38.6k lines, 1,044 passing tests, and every performance number measured on real hardware. |
| CodeGen — Autonomous CP Solver | The end-to-end AI solver I built for the Meta Hacker Cup 2025 AI Track, where it placed Global Rank 10. FastAPI + Gemini, multimodal problem-statement parsing, generate–test–repair loops. (dev history and client tooling) |
| BidRAG | Open-source RAG extraction for RFQs, tenders and contracts, with every answer traceable back to its source page. An open reimplementation of Bidify, the internal Siemens application I worked on — same feature set, lighter, and swappable at every layer (LLM, embeddings, extraction, storage, auth) with a free offline default everywhere. |
| Olist Marketplace Intelligence | An end-to-end analyst project on Brazilian e-commerce data. Late delivery causes a 1.70-point drop in review score (95% CI [−1.76, −1.67]), survives four robustness attacks including an E-value of 11.17, and the failing leg is carrier transit, not the seller. Tested dbt/DuckDB warehouse, runs in ~90s from a clean clone. Live Tableau dashboard |
| Context-Aware Multimodal Retrieval | A multimodal RAG pipeline over PDFs that retrieves across text, tables, figures and formulas with cited answers. Modality-specific pipelines pick the right LLM per content type; three retrieval paths run in parallel (summary embeddings, raw atomic content, CLIP text-to-image). |
| Non-conflicting Timetable Generator | University timetabling as a constraint satisfaction problem in OR-Tools CP-SAT — nine lexicographic objective tiers, so "why was my morning preference broken?" is mechanically answerable. Verified by an independent checker that shares no code with the solver, and it diagnoses ("you are 108 teaching hours short") instead of saying infeasible. |
| SENTRAL | Multi-spectrum equity analysis: 14 fundamental metrics (with Piotroski F-Score and Altman Z-Score), 18 technical indicators, news sentiment from 10 LLMs over a corpus gathered from seven APIs, and Transformer/LSTM forecasting, ensembled into buy/hold/sell probabilities. |
| LitigatAI | An AI litigation and dispute-resolution platform for Indian legal practice, with 10 integrated tools and a live demo. |
| TimeWeave | A team hackathon project for American Express (I wrote the bulk of it): cross-channel journey stitching into one identity, churn scoring with per-driver attribution in raw feature space, and counterfactual cohorts — every recommendation labelled as correlational, never as causal uplift. (team repo) |
More projects - legal NLP, medical imaging, federated learning, CV and tools
| Project | What it is |
|---|---|
| Legal-RAG | A citation-grounded RAG chatbot over Indian statutes and case law, with hybrid vector + BM25 retrieval. |
| Consumer Commission Complaint Agent | A guided-intake agent that turns a plain-language grievance into a file-ready Indian Consumer Commission complaint. Live demo available. |
| Lung Cancer Detection with Soft Attention | Soft-attention deep transfer learning for lung cancer identification from CT scans and histopathology. Published at IEEE ISACC 2025. |
| Federated Cervical Cancer Detection | Cervical cancer detection on Pap smear images, comparing three federated learning strategies. |
| Federated Heart Disease Prediction | Predicting heart disease from clinical biomarkers under federated learning. |
| Face Liveliness Detection | A lightweight, fast anti-spoofing model that tells real faces from fake images and videos. |
| Gesture Video Controller | Webcam hand-tracking to control media playback with single-hand and dual-hand gestures. |
| Job-Aligner | Tailors a LaTeX resume to any job description and compiles a submission-ready PDF, with a Chrome extension front end. |
| Text-to-Handwriting Converter | Converts typed text (including tables) into realistic handwriting and exports it as an A4 PDF. |
