Phantom842

Project Intern

• •Completed a project-based work experience program focused on Adobe Express and industry-standard creative workflows.
• •Collaborated with a 4-member team to develop 5 projects and deliver 10 creative deliverables, incorporating feedback and iterative design improvements.

buildspace - Nights and Weekends Programme

Remote

• •Developed an acoustic fire suppression project exploring sound-based fire control, collaborating with a global builder community to iterate on the concept.
• •Design targets: 30-100 Hz at 100-120 dB, 10-30 cm range, 5-15 s suppression time, 20-30 test trials and a 60-80% target success rate on a small controlled flame.

Python, FastAPI, PostgreSQL, React

• •Built a bidirectional LLM security gateway that screens every prompt and response through 8 detectors across 3 tiers (unicode and encoding pre-processing, regex and YARA rules, ML classifiers), mapped to OWASP LLM Top 10 risks.
• •Added canary-token leak detection and a YAML policy engine (allow/sanitize/challenge/block), with Kafka event streaming, Docker deployment, a WebSocket incident console and precision/recall/F1 benchmarks on labelled attack and benign sets.

Python, PyTorch, Geospatial ML

• •Built a physics-constrained deep learning pipeline that reconstructs ocean temperature and salinity at 18 depths (0-2,000 m) from satellite surface data, trained on 8,340 Argo profiles matched to satellite passes (Indian Ocean, 2021-2023).
• •Validated on a held-out Bay of Bengal region (1,044 profiles) and a held-out monsoon season; uncertainty bands calibrate to 81% coverage at the 80% level and 95% at the 95% level.

Python, LLM Agents, Streamlit

• •Built a reconciliation agent that proves each payment-ledger mismatch with a sandboxed, re-executable script hashed into an append-only chain; on a 60-record test batch it scored precision and recall of 1.0 at about 3,700 records/sec.
• •Added a whole-batch conservation check that caught a systematic INR 175.87 ledger drift, plus a self-adversarial fuzzing layer that hunts the agent's own blind spots.

Full Stack Development, Cyber-Security

• •Built a cloud-based healthcare interoperability platform enabling secure real-time hospital data exchange using APIs.
• •Implemented role-based access control and a scalable patient identity system for secure inter-hospital coordination.

Python, FalkorDB, Graph Queries

• •Built a temporal knowledge graph from thousands of merged pull requests that shows who actually knows a module, why the code is the way it is, and how much unwritten knowledge leaves when a maintainer does.
• •Added what-if forks of the graph to simulate a contributor stepping away or a decision being reversed.