Akash Balamurugan

Co-Creator & Lead Engineer at Content Coach AI (Independent Project)

Built a Multi-tenant Agentic AI content creation platform where creators automate their writing workflow with customizable agents that draft, retrieve, and analyze posts in their own style; deployed on AWS. Architected the full-stack system end-to-end (React 19/Vite, FastAPI, PostgreSQL 18 + pgvector) with a LangGraph multi-agent backend. Built a multi-agent system where specialized agents collaborate to research, draft, analyze and write content via a shared tool layer (semantic search, websearch) with a human-in-the-loop approval step. Engineered a RAG pipeline (LangChain, pgvector, HNSW, Gemini embeddings) plus an adaptive style-memory engine so the AI writes in each user’s evolving voice; multi-tenant, CI-tested with GitHub Actions.

Co-Creator & Lead Engineer (Independent Project) at Content Coach AI

Built a multi-tenant agentic AI content creation platform deployed on AWS. Architected the full-stack system end-to-end (React 19/Vite, FastAPI, PostgreSQL 18 + pgvector) with a LangGraph multi-agent backend. Implemented a multi-agent system where specialized agents collaborate to research, draft, analyze, and write content via a shared tool layer (semantic search, websearch) with a human-in-the-loop approval step. Engineered a RAG pipeline (LangChain, pgvector, HNSW, Gemini embeddings) plus adaptive style-memory engine to preserve evolving user voice; ensured multi-tenant architecture with CI testing via GitHub Actions.

Co-Creator & Lead Engineer at Independent Work Content Coach AI

Built a multi-tenant agentic AI content creation platform deployed on AWS. Architected end-to-end full-stack system (React 19/Vite, FastAPI, PostgreSQL 18 + pgvector) with a LangGraph multi-agent backend. Built a multi-agent system where specialized agents collaborate to research, draft, analyze and write content via a shared tool layer (semantic search, websearch) with a human-in-the-loop approval step. Engineered a RAG pipeline (LangChain, pgvector, HNSW, Gemini embeddings) plus adaptive style-memory engines so the AI writes in each user’s evolving voice; multi-tenant, CI-tested with GitHub Actions.

AI/ML Engineer Intern at Inria

Developed a ML-based Deep Reinforcement Learning scheduler for proactive scaling for microservice deployment across edge-cloud infrastructure, reducing end-to-end latency by 70% using PyTorch. Engineered high-volume ETL data pipelines ingesting 50+ distributed edge-cloud telemetry system metrics using Pandas and NumPy for live model training. Designed optimization-oriented workload deployment learning logic for the ML model, balancing latency, compute resource availability, and network constraints across distributed infrastructure improving deployment stability.

AI/ML Engineer Intern at Inria (Institut national de recherche en sciences et technologies du numérique)

Developed a ML-based Deep Reinforcement Learning scheduler for proactive scaling for microservice deployment across edge-cloud infrastructure, reducing end-to-end latency by 70% using PyTorch. Engineered high-volume ETL data pipelines ingesting 50+ distributed edge-cloud telemetry system metrics using Pandas and NumPy for live model training. Designed optimization-oriented workload deployment learning logic balancing latency, compute resource availability, and network constraints across distributed infrastructure, improving deployment stability.

Software Engineer at Cognizant Technology Solutions

Developed Python (FastAPI) based microservices for payments and logistics modules powering PepsiCo North America’s FLNA enterprise supply-chain platform. Optimized PostgreSQL query logic to resolve concurrency bottlenecks, improving end-to-end responsiveness. Improved system efficiency by 30% by re-modeling the logic and API workflows to remove bottlenecks and strengthen concurrency, scalability, and reliability. Built a Python automation tool for enterprise PDF ticket validation, reducing manual effort by 90% with 100% audit accuracy.