Mukhil Baskaran

About Me

Since childhood, I've been fascinated by the inner workings of computers. I was that kid who dismantled desktops just to peer at the circuitry beneath, driven by curiosity and a desire to understand how these machines functioned. This early passion for technology steered me towards a career in computer science. I recently graduated from Georgia State University with a degree in Computer Science. Currently, I'm working as a Founding Engineer at originERP, where I lead the architecture and development of enterprise AI SaaS platforms. My expertise primarily lies in technologies such as Python, TypeScript, React, Next.js, FastAPI, PostgreSQL, and AI/ML frameworks.

When I'm not coding, I indulge in my hobbies that include learning about space, cooking, and spending quality time with friends and family. I am constantly seeking to learn new skills and dive into diverse fields, enriching both my personal and professional life.

My experience

Founding Engineer

originERP

San Francisco, CA (Remote)

• Solo-architected enterprise AI SaaS from first commit to production alpha (10 users) in 5 months as sole technical owner reporting to investors, designed 16-table multi-tenant PostgreSQL schema with RLS, integrated four AI providers (OpenAI, Anthropic Claude, Cohere, Gemini), shipped conversational agent system with streaming inference, and consolidated ERP blueprint creation into a single platform where alpha users reported significant time savings over traditional multi-tool consultant workflows.
• Shipped split-pane chat + collaborative document editor where agents generate reports post-consultation and users directly edit them in TipTap with real-time collaboration, built FastAPI backend with typed Python service layers and Next.js 15/React 19 frontend with robust JWT session management, evaluated RAG pipeline achieving Recall@5 of 0.83 (human-labeled) at <500ms P95 latency using Qdrant + Cohere, built deterministic validation layers and tooling to harden AI outputs for production readiness.
• Stood up three-tier environment strategy (dev, staging, prod) and migrated from bare metal (PM2) to containerized Docker deployments with GitHub Container Registry, Coolify Cloud, and trunk-based promotion workflow as team expanded, owned deployment infrastructure on DigitalOcean with Nginx, Docker and GitHub Actions CI/CD.

Feb 2025 - Present

Freelance Software Engineer

EMeRG

Calgary, Canada

• Contracted directly by business client to extract data from 200+ hospital websites, delivered AI-powered web scraper achieving 80% extraction accuracy in 2-month timeline using Crawl4AI + Groq AI, handling requirement gathering, architecture decisions, and iterative client feedback.
• Shipped React frontend with PostgreSQL backend featuring validation pipelines, structured storage, and retry logic, supporting repeatable on-demand extraction workflow for healthcare data processing, deployed on Vercel with GitHub Actions CI/CD.

Nov 2024 – Jan 2025

Student Software Assistant

Panther Dining - Chick-fil-A

Atlanta, GA

• Eliminated manual paper-based inventory tracking for campus Chick-fil-A by building full-stack system adopted by 10+ staff for managing 200+ SKUs daily, built production REST API with PostgreSQL, optimized queries, JWT auth, and Python backend + Next.js frontend, improving operational efficiency store-wide.
• Deployed on AWS EC2 and maintained production system over 2 years while balancing full-time coursework, handled continuous bug fixes, performance optimization, and feature requests, keeping system stable for daily store operations.

Sep 2022 – Oct 2024

My projects

FinsightAI
Full-stack AI-powered expense tracker using OpenAI (spending insights), OAuth 2.0, and encrypted storage; organically adopted by 10+ friends tracking 200+ transactions and deployed on Vercel with GitHub Actions CI/CD and a dockerized stack.
- React
- Next.js
- TailwindCSS
- AI
Sentinel RAG
IaC security scanning and autofix assistant with per-user sandboxes, a chat UI, and a patch/diff workflow. Runs real scanners (Checkov, tflint), proposes fixes via a constrained apply_patch tool with reviewable diffs, and operates in isolated per-session sandboxes.
- Python
- FastAPI
- Next.js
- OpenAI
TinyTasks - Voice-First AI Task Management App
Pivoted capstone project in 2 weeks after user feedback—integrated Groq Whisper for voice recognition and Llama models for intelligent task extraction, transforming a basic task app into a voice-first AI experience with automatic subtask generation. Shipped a cross-platform mobile app (iOS/Android) with a Flask AI backend, React Native/Expo frontend, and real-time Supabase calendar sync; dockerized services for deployment—reducing task entry from manual typing to 5-second voice commands with automatic structuring.
- React Native
- Expo
- Node.js
- Flask