May 2026 B.Tech IT graduate (CGPA 8.62, Best Outgoing Student — IT Dept.) | Chennai, India

Building flight-grade software at the intersection of:

1. Fault-tolerant aerospace avionics & autonomous systems
2. Physics-informed ML for orbital mechanics / telemetry
3. Quantum-secure satellite communications

📅 Currently: Studying Japanese daily · Preparing MEXT / OIST Spring 2027 applications for Japan · Actively reaching out to Japanese university labs and space/AI/embedded firms for funded Master's positions and engineering roles starting 2027. Open to short remote contract/trial engagements to prove value first.

• 🚀 ROXα fault-tolerant rocket flight computer — awarded "Excellent" by ISRO / IN-SPACe / ASI, ranked top 0.8% out of 73,000+ national participants
• 🧠 SMART-DEORBIT PINN orbit propagator — <1 km altitude error under J2 + NRLMSISE-00 drag, 100× speedup over DOP853 numerical solvers; paper accepted and presented at MAHSA University international conference
• ⚛️ qkdpy — open-source QKD library on PyPI (v0.3.0, 17 releases, 200+ unit tests; BB84 / E91 / CV-QKD protocols)
• 🛰️ Telemetry Anomaly Agent — LLM-assisted anomaly detection on ESA-AD 76-channel satellite telemetry, production FastAPI/Next.js deployment with TTL cache and backoff/fallback
• 🎓 ASPIRE Leadership Program (Harvard initiative) graduate, Extended Leadership Pathway; President, RISE AI/ML Club; Core Member, Dr. MGR ACS Space Tech Centre.

ISRO/IN-SPACe/ASI "Excellent" award, top 0.8% of 73,000+ teams. High-reliability firmware on Teensy 4.1 featuring a 13-state BOOT→SAFE finite state machine, dual-redundant sensor fusion (ICM-20948 + BMI088 IMUs, MS5611 + BMP280 barometers, dual QMC5883L compasses) with voting logic, CRC-framed XBee S2C telemetry downlink, independent watchdog fault containment, and a Python ground control station with real-time telemetry parsing and alerting.

Built under vibration and hard 5 ms real-time control-loop constraints.

C++ Teensy 4.1 Kalman Filtering Sensor Fusion XBee Radio I2C/SPI Python GCS

Physics-informed neural network for rocket stage de-orbit prediction, incorporating J2 gravitational perturbations and NRLMSISE-00 atmospheric drag. Achieves <1 km altitude error and 100× speedup vs adaptive DOP853 numerical propagators — fast enough for near-real-time conjunction assessment. Paper presented at MAHSA University international engineering conference (Malaysia).

Python PyTorch SciPy Orbital Mechanics PINNs

Production-ready open-source Python library for quantum key distribution simulation, released on PyPI (v0.3.0, 17 releases, 200+ unit tests, 5 GitHub stars). Implements BB84 (prepare-and-measure), E91 (entanglement-based), and CV-QKD continuous-variable protocols with realistic noise/error models, designed for research and teaching in quantum-secure communications — including over optical inter-satellite links.

Python Quantum Cryptography BB84 E91 CV-QKD PyPI Packaging

End-to-end spacecraft mission-ops agent: Isolation Forest for unsupervised anomaly detection on the ESA Anomaly Dataset (76 telemetry channels), Gemini 2.5 LLM for root-cause diagnostic report generation, deployed as a FastAPI microservice with Next.js dashboard, TTL response caching, and LLM API backoff/fallback for reliable ops use.

Python Scikit-Learn FastAPI Next.js Google GenAI SDK Anomaly Detection

• Email is the fastest way to reach me: pranavakumar.it@gmail.com — I reply within 24 hours.

Code that works on the first try on real hardware beats code that looks perfect on a whiteboard.