Thanks to visit codestin.com
Credit goes to github.com

Skip to content
View pmg5408's full-sized avatar

Block or report pmg5408

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.

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
pmg5408/README.md

Hi, I'm Pratham (Prat) 👋

Software engineer building end-to-end full-stack and AI systems in production. Strong CS fundamentals, as well as AI proficiency — I maintain complete ownership, know how to navigate ambiguity and like operating close to the stakeholder.

📍 DC area · 🛰️ Hughes Network Systems (EchoStar)


🚀 What I'm building right now

An agentic anomaly-detection system for the Hughes satellite network — end-to-end, three moving parts:

  • Ingestion framework — a source-agnostic pipeline built for scale from day one: orchestration fully decoupled from the data, with a clean seam where each new source plugs in its own processing without touching the core.
  • Detection — Parquet feeds an in-house ML platform over an API to score anomalies.
  • LLM reasoning layer — takes the model output plus business context, pulls raw data and trends as needed through its own tools, then catches missed anomalies, flags false positives, and attaches confidence scores the systems team acts on.

🛠️ Selected work

RGB Collision Guardrail → DimOS
open-source contribution to a 3.6k⭐ agentic-robotics OS
A robot motion-safety guardrail designed as cleanly separated runtime, policy, and state layers — so it drops into any skill built on the OS, and the policy is swappable without touching the rest. automated review scored it 5/5.

chat-context
MCP memory server · Python · ChromaDB · OpenAI
Persistent semantic memory for LLM chats. Cut API cost ~40% by retrieving only relevant context instead of replaying full history.

grocery-meal-agent
FastAPI · React · Postgres · Celery · Redis · Docker · AWS
Full-stack AI assisted meal planner with async task processing — built front to back.


🧰 Stack

Languages · Python · TypeScript · C++
AI & Agents · Multi-Agent Systems · RAG · Vector DBs · MCP · LangGraph
Backend · FastAPI · Celery · Redis · Kafka · WebSockets
Infra & Data · Kubernetes · Docker · PostgreSQL · MongoDB · AWS · Linux


🔗 Find me

🌐 v0-pratgala.vercel.app · 💼 LinkedIn
📧 [email protected] · 📱 +1 (617) 751-9429

Pinned Loading

  1. grocery-meal-agent grocery-meal-agent Public

    Python 1

  2. chat-context chat-context Public

    Python 2

  3. dimos-collision-guardrail dimos-collision-guardrail Public

    Python

  4. Linear-Memory-Storage-Device Linear-Memory-Storage-Device Public

    C