🚀 CSE Undergraduate @ NIT Silchar
💻 Systems Engineering • Backend • Quant
⚡ Building high-performance systems & intelligent applications
- 🔧 I love low-level systems + performance optimization
- 📊 Exploring Backend Engineering & quantitative finance
- 🤖 Interested in agentic AI & real-time analytics
- 🏆 Competitive programmer & open-source contributor
📌 FastAPI, Go, Docker, AWS EC2, Postgres RDS
- Designed and deployed a distributed trading-engine evaluation platform on AWS, supporting automated correctness validation and large-scale performance benchmarking.
- Built a secure containerized execution environment using Docker and gVisor for safe execution of untrusted participant code.
- Developed a distributed Go benchmarking framework capable of generating realistic market traffic and sustaining 1.3K+ TPS under 1,000 concurrent traders.
- Implemented a FastAPI orchestration service for worker coordination, benchmark scheduling, metrics aggregation, and automated scoring.
- Created a real-time competitive leaderboard system using PostgreSQL and WebSockets for live ranking and benchmark analytics.
📌 FastAPI, CatBoost, OSMnx, NetworkX, AWS, Mappls APIs
- Built a production-ready ML-driven traffic intelligence system for Bengaluru using 8K+ traffic incidents, enabling congestion prediction, incident impact estimation, and intelligent diversion planning at city scale.
- Architected a modular multi-stage inference pipeline: feature engineering → closure probability prediction → impact scoring → response recommendation → diversion routing.
- Optimized city-scale routing by partitioning Bengaluru into 56 OSM road graphs with two-level lazy loading, reducing active memory usage by 94% (4.8 GB → 280 MB) while achieving sub-150 ms warm route generation.
- Implemented diversion route search using KD-tree spatial indexing + Dijkstra's algorithm, and validated selected routes using Mappls traffic-aware APIs for real-time ETA and distance estimation.
- Built closed-loop feedback system, capturing post-incident outcomes for periodic model recalibration and continuous improvement in prediction accuracy.
🔥 TitanDB – High-Performance KV Store Repo
📌 C++, Systems Programming
- ⚡ O(1) lookups using in-memory hash index
- 💾 Disk-backed storage with fixed-width records
- 🧱 Inspired by Bitcask architecture
- 🔄 Log compaction + crash recovery
🌐 TrafficTalk – AI Network Monitoring Assistant Repo
📌 Python, Redis, LLMs
- 📊 Real-time network traffic analysis
- ⚡ Sub-10ms processing latency
- 🤖 LLM-powered insights (Groq + LLaMA 3)
- 🔄 Hybrid agent system
-
🔧 CGAL
- Fixed cross-platform build issues (CMake, toolchain)
- Enforced C++17 compatibility
-
🌍 DBPedia
- Built Human-in-the-Loop dashboard
- Improved Hindi knowledge graph
- 🥈 2nd – Arbitrage Arena (IISc)
- 🥉 3rd – IISc Honour Code Hackathon
- ⚡ Meta HackerCup Rank: 2245
- 🏅 Top Contributor – Open Source Marathon (NIT Silchar)
- 🚀 High-performance backend systems, distributed systems
- 📊 Quant + ML systems
- 📧 Email: abhigyanph@gmail.com
- 🧠 Codeforces / LeetCode
I enjoy building systems where latency matters more than lines of code ⚡