Saksham Kumar
LLM / GenAI Engineer
New Delhi, India
Building Full Stack Applications, LLM Application systems at scale.
About
I'm a Software Engineer focused on building scalable, AI‑powered full‑stack applications.
I'm currently a founding engineer at Poplar AI — building an AI social media scheduling platform. I'm also learning Rust and Solana. I'm also building khojai.in — an AI job automation tool that turns job hunting into a conversation.
Recently, I built the official scoring platform for international Yogasana championships at YUGA (JoinYuga) and developed AI agents for B2B travel workflows at ChatDMC.
Writing & Highlights
Stack
Experience
Founding Full‑Stack AI Engineer (Remote)
Full-time/Contract·
- Building an AI-powered social media scheduling platform as the first engineering hire (US-based, remote).
- Designed and delivered end-to-end features across Next.js app, Supabase/Postgres, and background agents with Inngest.
- Integrated OAuth and official APIs for Facebook, Instagram, LinkedIn; implemented media storage on AWS S3.
- Implemented LangGraph-based agents for scheduling/automation; payments via Polar.
Founding Engineer
Next.js
TypeScript
Supabase
LangGraph
OAuth
Facebook API
Instagram API
LinkedIn API
AWS S3
Inngest
Polar Payments
Remote
US-based
Full-Stack Developer
Internship·
- Championship Platform Development: Built the official scoring platform for the 2nd Asian Yogasana Sports Championship from scratch using Next.js, serving 16+ countries and 250+ organizations.
- Frontend Architecture: Developed comprehensive UI from Figma designs including authentication, stores with Zustand, and optimized user interfaces for judges and athletes.
- Real-time Scoring System: Created live scoring platform used by judges across multiple sections with real-time synchronization and cross-screen functionality.
- Team Collaboration: Led core development team and mentored later interns, handling fast-paced development and real-time problem solving during live deployments.
- Backend Integration: Contributed to backend development ensuring rock-solid performance for international championship events.
- Platform Features: Built championship management, athlete portfolios, rating systems, and analytics dashboards for the Yogasana Universal Grading & Analytics platform.
- Live Event Support: Provided on-site technical support during international championships, handling real-time issues and ensuring smooth platform operation.
Next.js
TypeScript
Zustand
Real-time
UI/UX
AI Agent Developer
Contract·
- AI Agent Development: Built end-to-end AI agent called 'Rate DMC' for B2B travel platform using LangGraph and LlamaIndex for intelligent query processing.
- Full-Stack Implementation: Developed complete frontend to backend solution in Next.js with exact UI implementation from Figma designs.
- Intelligent Query Processing: Created AI chatbot that retrieves data, makes rate queries, raises queries, and generates quotes for travel agents.
- B2B Travel Platform: Built AI agent for ChatDMC platform connecting travel agents with Destination Management Companies (DMCs) worldwide.
- Real-time Communication: Implemented chat-based interface for instant communication between travel agents and DMCs.
- Document Management: Integrated document uploads, payment management, and booking confirmations within the AI agent system.
- Multi-Quote System: Developed system allowing travel agents to submit queries and receive competitive quotes from multiple DMCs.
- Platform Integration: Seamlessly integrated AI agent with existing ChatDMC platform serving 50+ travel agencies.
TypeScript
Python
Node.js
React
Next.js
PostgreSQL
tRPC
Express.js
OpenAI
LLM Orchestration
Prompt Engineering
Docker
Azure
Education
Projects(7)
- Architected an autonomous, multi-intent AI agent using LangGraph to manage the entire job application lifecycle, from discovery to post-application tracking.
- Engineered a full-cycle application automation system that identifies actionable jobs (e.g., LinkedIn), queues them in a database, and hands off tasks to a browser extension via secure REST APIs for 'one-click apply' execution.
- Developed an intelligent application tracking module that uses the Gmail API to scan for replies from companies, leveraging a classification LLM to categorize email responses into statuses like 'Interview Scheduled' or 'Rejected'.
- Built a multi-stage contact enrichment pipeline using Exa for LinkedIn profile discovery and the Prospeo API for email verification to find key decision-makers at target companies.
- Expanded the real-time data ingestion pipeline to over 20 sources, adding LinkedIn Jobs API and integrating advanced browser automation with Hyperbrowser/Puppeteer for complex data extraction.
LangGraph
Node.js
TypeScript
Express.js
Microservices
Google Gemini
Groq
OpenAI
Supabase
PostgreSQL
Redis
AWS S3
Docker
Kubernetes
Gmail API
Exa API
Prospeo API
Puppeteer
Hyperbrowser SDK
- GitMind indexes your repositories, connects the dots, and provides instant, accurate answers to your code-related questions.
RAG
LangChain
Vercel AI SDK
tRPC
Next.js
Tailwind CSS
Prisma ORM
FireBase
PostgreSQL
Next Auth
Clerk
Octokit