Binayak Bhattacharjee

- Contributing to backend feature development using Node.js, Express.js, and REST APIs, collaborating on scalable backend services and data models in an Agile environment. - Working on Retrieval-Augmented Generation (RAG) systems and agentic AI workflows, gaining hands-on experience with backend AI integration and production-ready architectures.

- Built and deployed smart contracts using Rust on the Internet Computer Protocol (ICP) with dfx. - Developed and tested backend logic for DApps in a Linux (Ubuntu) environment. - Worked with canisters, ICP identity, and secure data storage models.

Distributed Job Scheduler

June 2026

- Built a distributed job scheduling platform using Spring Boot, Quartz Scheduler, and PostgreSQL, supporting cron-based task execution and persistent scheduling. - Designed and implemented RESTful APIs with Spring Data JPA/Hibernate, PostgreSQL-backed Quartz persistence, and Redis integration for distributed coordination and caching. - Engineered a fault-tolerant scheduling architecture where Quartz triggers persist in PostgreSQL and dispatch jobs through Apache Kafka, enabling reliable recovery, asynchronous task processing, and scalable execution.

Research Paper RAG Pipeline

March 2026

- Built an end-to-end Retrieval-Augmented Generation (RAG) system for semantic search and question answering over research papers using hybrid retrieval (dense + sparse) techniques. - Implemented FAISS vector indexing, BM25 keyword retrieval, and cross-encoder reranking, improving retrieval reliability by reducing failure rate from 33% to 0% on a custom evaluation benchmark. - Designed an evaluation framework with Recall@K, MRR, and failure analysis, and integrated LLM-based grounded answer generation with citation-aware context building.

Low-Latency Telemetry Engine

December 2025

- Engineered a high-throughput sidecar logging system in C++ using Shared Memory IPC (mmap) to decouple application logic from disk I/O, enabling a real-time telemetry data ingestion pipeline and eliminating logging pauses in latency-sensitive environments. - Implemented a lock-free Single-Producer-Single-Consumer (SPSC) ring buffer using std::atomic and custom memory barriers, ensuring thread safety without Mutex overhead or system call context switching. - Achieved sub-microsecond write latency (P99) and a sustained throughput of 2 million logs/second, validated against standard std::ofstream using a custom Python benchmarking suite on real-world crypto trade datasets.