Skip to content
View Amitesh-Sinha5's full-sized avatar

Highlights

  • Pro

Block or report Amitesh-Sinha5

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Amitesh-Sinha5/README.md

Amitesh Sinha

CS undergrad at PES University. I build distributed systems and like the parts that only show up under failure — consensus, replication, and what a cluster does when a node stops answering.


Selected work

mini-raft — Raft from scratch in Node.js Randomized election timeouts, term-based voting with the §5.4.1 log up-to-date restriction, AppendEntries log repair with conflict-term backtracking, and majority-quorum commit across four replicas — under a real-time collaborative canvas. Includes a partition simulator for reproducing split-brain. In-memory state; the README is explicit about which paper guarantees that costs. Node.js · Docker · WebSockets

mini-hdfs — distributed file store Namenode with fsimage + append-only edit log and periodic checkpoint merge, 2 MB chunking with SHA-256 checksums, replication factor 2 with heartbeat-driven failure detection and automatic re-replication, free-space-aware placement. Python · TCP sockets · threading

helix-agentic-ai-ops — agentic AI platform Three LangGraph agent pods on one governed backend: auth and RBAC, tier-aware rate limiting, hybrid retrieval with RRF and reranking, semantic caching (p95 38.5 ms → 10.7 ms under a 5,106-request load test), cost telemetry, and a CI gate on answer quality. Runs fully offline on a deterministic mock LLM. FastAPI · PostgreSQL · Redis · Terraform · ECS/K8s

NeuroFit — real-time pose analysis with an LLM safety layer Camera-based strength-training assistant built to test whether grounded adversarial self-reflection reduces unsafe coaching advice. Python · computer vision

Structural-Coder — collaboration, top contributor (31 of 55 commits)

HoneyTrap Firewall — packet filtering, IP banning, and attacker logging with a honeypot adapter layer. Python


Currently

Reading the Raft and Dynamo papers with an editor open, and looking for a systems-side internship where storage and replication are the product rather than a dependency.

sinhaamitesh0@gmail.com

Pinned Loading

  1. helix-agentic-ai-ops helix-agentic-ai-ops Public

    Agentic AI operations platform: FastAPI, PostgreSQL, Redis, hybrid retrieval with RRF + rerank, semantic caching (p95 38.5ms uncached to 10.7ms cached under a 5,106-request load test), 145 tests ac…

    Python

  2. mini-hdfs mini-hdfs Public

    Mini-HDFS in Python: namenode with fsimage + append-only edit log and periodic checkpoint merge, 2MB chunking with SHA-256 checksums, replication factor 2 with heartbeat-driven failure detection an…

    Python

  3. miniraft miniraft Public

    Raft consensus from scratch in Node.js: randomized election timeouts, term-based voting with the log up-to-date restriction, AppendEntries log repair with conflict-term backtracking, and majority-q…

    JavaScript

  4. NeuroFit NeuroFit Public

    Real-time camera-based strength-training assistant with a self-reflective safety layer over its LLM coaching, testing whether grounded adversarial self-reflection reduces unsafe advice.

    Python

  5. mlcommons/GaNDLF mlcommons/GaNDLF Public

    A generalizable application framework for segmentation, regression, and classification using PyTorch

    Python 191 91

  6. RahulAnand2077/Structural-Coder RahulAnand2077/Structural-Coder Public

    A Smart PyTorch Code Generator powered by a Knowledge Graph + Heterogeneous GNN

    Jupyter Notebook