Available for Freelance AI Contracts & MVPs

Kumar Harsh

Freelance AI Engineer & Systems Architect

I architect autonomous AI agents, multi-agent swarms, custom RAG pipelines, and high-performance intelligent apps that ship in days—not quarters. Based in Bengaluru, India · Shipping for founders & high-growth teams worldwide.

Estimate Project Scope ⚡ Explore Shipped Apps (9) ↓ GitHub (68 Repos)
Flagship AI App Active Prod

Memory 🧠

Cognitive internet second brain with Gemini semantic search and URL context retention.

Turnaround Speed
1–2 Weeks

Zero-to-One Rapid AI MVP Sprint

Freelance Edge

Full-stack AI. Not just prompt scripts in notebooks, but production systems with streaming UX and vector infrastructure.

Core Production Stack ENTERPRISE-PROVEN
Python Gemini 2.5 / OpenAI LangChain / Agents React 19 / TypeScript Vector DBs / RAG Cloudflare Workers & D1
Services & Capabilities

How We Can Work Together

High-velocity, production-grade freelance engineering for founders, venture-backed startups, and modern engineering teams.

Estimate Your Project Scope ↗
01 / AGENTS

Autonomous AI Agents & Swarms

Transform manual operations into self-directed AI workflows. Multi-step reasoning, external tool-calling, browser actions, and reliable execution.

  • Multi-agent planning & execution swarms
  • Custom tool calling (Slack, CRM, SQL, APIs)
  • Human-in-the-loop review guardrails
  • Token budget limits & observability logging
TYPICAL TIMELINE: 2–3 WEEKS Select ⚡
02 / SPRINT

Rapid 1–2 Week AI MVP Sprint

Zero-to-One: Turn your idea into a production-ready, beautifully designed web app. Complete with streaming LLM UX, auth, database, and payments.

  • Full-stack React 19 + TypeScript + Node architecture
  • Streaming AI chat/generation interfaces (zero latency anxiety)
  • User authentication, database models & payment hooks
  • Cloud deployment with CI/CD ready on Day 14
TYPICAL TIMELINE: 10–14 DAYS Select ⚡
03 / RAG & CONTEXT

Enterprise RAG & Context Memory

Ground your AI in private documents, knowledge bases, and user session context. Modeled after our proprietary Memory architecture.

  • High-recall vector search & hybrid semantic retrieval
  • Dynamic document chunking & contextual re-ranking
  • Multi-turn session memory retention across visits
  • Rigorous anti-hallucination verification checks
TYPICAL TIMELINE: 2–3 WEEKS Select ⚡
04 / RESILIENCE

AI Reliability & Chaos Engineering

Eliminate rate limit crashes, model downtime, and flaky socket drops. Stress-tested with battle-hardened Hiccup chaos engineering.

  • Intelligent multi-model fallback cascades (Gemini/OpenAI/Anthropic)
  • Exponential backoff retry budgets & circuit breakers
  • Flaky connection simulation & chaos testing suite
  • Sub-500ms TTFT (Time-to-First-Token) latency tuning
TYPICAL TIMELINE: 1–2 WEEKS Select ⚡
About Harsh
"I bridge the gap between cutting-edge LLM research and production software: building reliable autonomous agents, low-latency streaming interfaces, and resilient data pipelines that founders can scale."

With over 4+ years of professional engineering experience, Kumar Harsh is an elite Freelance AI Engineer & Systems Architect based in Bengaluru, India. He partners with founders, high-growth startups, and visionary teams globally to take AI products from napkin concepts to production reality.

Kumar's engineering foundation was forged at enterprise scale: architecting AI agent pipelines at Allianz Technology, leading generative AI R&D for engineering automation at Dassault Systèmes, and building IoT software at UST. Today, his focus is rapid zero-to-one execution—combining deep LLM reasoning, vector memory layers, and buttery-smooth React 19 interfaces.

9 Live Apps Shipped (Last 6 Mo)
1–2W Average MVP Sprint Turnaround
68+ Open Source GitHub Repos
Core Competencies

AI Systems & LLM Architecture

Autonomous AI Agents Multi-Agent Swarms Vector Databases & RAG Function & Tool Calling Multi-Model Fallbacks Automated LLM Evals Gemini 2.5 Flash / Pro OpenAI GPT-4o

Full-Stack & Cloud Infrastructure

React 19 & Next.js TypeScript Node.js & Express Python & FastAPI Cloudflare Workers & D1 Tailwind CSS Streaming WebSocket / SSE
Shipped Portfolio 03

Shipped in the Last 6 Months 🚀

Real, live, tested applications built and shipped to production. No mock-ups, no placeholders.

9 Live Apps
01 / FLAGSHIP AI Shipped in Last 6 Months · Active Prod

Memory 🧠 — The Internet's Second Brain

A persistent cognitive layer for the web. Attach thoughts, notes, and task references to any URL or domain, and watch them automatically resurface when you revisit pages. Built with Gemini semantic vector recall, zero-shot classification, and interactive browser overlays.

Gemini 2.5 API React 19 Vector Memory Express Backend TypeScript
Live App ↗ GitHub Repo
02 / AUTONOMOUS WHATSAPP AGENT
₹0/Mo Serverless Stack Cloudflare + Meta + Gemini

Honey 🍯 — Autonomous WhatsApp AI Assistant & Life OS

A private, witty, always-on AI assistant residing directly in WhatsApp. It remembers context and personal facts, sets natural language reminders, tracks expenses via camera receipt snaps, ghostwrites and schedules LinkedIn posts, processes voice notes across English/Hindi/Hinglish, and replies with synthesized audio. Multi-provider AI fallback cascade running serverless at ₹0/mo cloud cost.

Production Serverless Topology ₹0/mo cloud spend · Free-tier infrastructure
WhatsApp (User) ──► Meta Cloud API ──► Cloudflare Edge Worker (Global)
├──► AI Failover: Workers AI (Llama 3) ──► Groq ──► Gemini 2.5 Flash
├──► Cloudflare D1 Vector Database (Semantic Multilingual Embeddings)
└──► Minutely Cron Engine: 8 AM Briefing · 1 PM Quiz · 8:30 PM Habits · Market Alerts
🎙️ Voice Notes & Audio

Speak in English, Hindi, or Hinglish. Honey processes intents and replies with natural text plus audio voice notes.

🧾 Receipt & PDF OCR

Snap rupee bills to auto-log categorized expenses. PDFs are digested with extracted due dates scheduled automatically.

🧠 Semantic Recall

Search notes, facts, and journals by conceptual meaning using vector embeddings rather than rigid exact keyword matches.

💼 LinkedIn Post Factory

Turns quick voice brain-dumps into 5 ghostwritten posts in your authentic style, scheduled for Tue/Wed/Thu mornings.

📈 Market Alerts & News

15-minute price monitors for Indian equities (Reliance, Nifty, Sensex), Mutual Fund NAVs, and morning news digests.

⚡ Multi-LLM Resilience

Cloudflare Workers AI cascades into Groq, then Gemini 2.5 Flash if rate limits or provider slowdowns occur.

Cloudflare Workers Meta WhatsApp Cloud API Cloudflare D1 (SQLite) Gemini 2.5 Flash Groq LLaMA-3 Cron Engine ₹0/Mo Running Cost
View Honey on GitHub ↗ Commission a Custom WhatsApp Agent ⚡
03 / AI COPILOT Live Demo

Quick Reply AI (ReplyBro) 🤖

Conversational wingman that ends "what do I even say back to this?". Generates context-aware, witty, or razor-sharp professional responses tailored to intent, subtext, and platform dynamics.

LLM Orchestration LangChain TypeScript
04 / MULTIMODAL AI Live Demo

MemeGenius AI 🎭

Multimodal AI meme generation engine that grasps internet culture, tone, and humor vibe. Analyzes images and context via Gemini Vision and renders dynamic typography on custom canvases.

Gemini Vision Multimodal AI Generative Canvas
05 / CHAOS DEVTOOLS Live Tool

Hiccup ⚡ Network Chaos Sandbox

Gives developers and AI pipelines an endpoint that misbehaves—slow latency, random 500s, dropped sockets, and jitter. Test retry budgets and exponential backoff without building flaky mock servers.

Chaos Engineering API Resilience Edge Functions
06 / FINTECH UTILITY Live Tool

Invoicely 🧾 60s GST Invoicing

Lightning-fast, privacy-first GST invoice & bill-of-supply generator with dynamic NPCI-compliant UPI payment QR codes. Zero signups or walls—built for Indian freelancers and agencies.

FinTech NPCI UPI Engine React 19
07 / TELEMETRY STUDIO Live App

PACE 🏃 Telemetry Studio

Transform running, cycling, and fitness telemetry into publication-grade visual records and clean photo overlays inspired by Strava, Oura, and Scandinavian editorial design.

React 19 SVG Telemetry Motion Engine
08 / MOBILITY UI Live Demo

SmoothRide 🚗 Urban Mobility

Next-generation mobility & ride scheduling platform with buttery liquid micro-interactions, simulated route physics, and responsive ergonomics designed for instantaneous booking flows.

React Liquid Motion Interactive UI
09 / VISION ENGINE Live Tool

AsciiMe 🎨 Camera ASCII Engine

High-performance media converter and live camera streaming engine that converts raw pixels into real-time, customizable ASCII art with dynamic luminance matrix mapping.

TypeScript Canvas API ASCII Vision

Curious about earlier builds, experimental algorithms, or open-source libraries?

View All 68 Repositories on GitHub
Interactive Scope Estimator

Plan Your AI Build With Harsh

Select your project requirements below to see an instant timeline estimate, deliverables checklist, and launch a tailored inquiry directly to Kumar Harsh.

1 Select Project Scope
Autonomous AI Agent
Multi-tool reasoning, function calling, external database integration
Full-Stack AI Web MVP
Zero-to-one production web app with streaming UI, auth & payments
Enterprise RAG & Memory
Vector embeddings, smart chunking, re-ranking & private data grounding
Reliability & Stress Testing
Eliminate flaky API calls, rate-limit crashes & latency bottlenecks
2 Desired Timeline
⚡ Expedited Sprint
10–14 Days (Rapid zero-to-one prototype or critical feature)
🎯 Standard Production
3–4 Weeks (Full enterprise-grade build, docs & hardening)
💼 Monthly AI Retainer
Ongoing Fractional AI Systems Architect & Continuous Shipping
3 Architecture Add-ons (Optional)
Streaming Liquid UI
Buttery fast React 19 real-time token streaming
Automated AI Evals
LLM-as-a-judge regression tests for prompt changes
Multi-Model Fallbacks
Automatic failover between Gemini, OpenAI & Anthropic
Estimated Scope & Deliverables
Autonomous AI Agent · 10–14 Days Sprint
Includes: ReAct agent loop, tool integration (APIs/DB), error guardrails, and cloud deployment.
Launch This Brief with Harsh 🚀 Direct response within 12 hours guaranteed
Career Timeline
Mar 2026 — Present

Allianz Technology

Senior AI Agent & Software Engineer

Architecting resilient AI Agent workflows, enterprise LLM orchestration, and automated ML pipelines for core global financial and insurance operations.

2022 — Mar 2026

Dassault Systèmes

AI Researcher & R&D Engineer

Led research into generative AI for industrial design. Built autonomous agents that can navigate complex engineering schemas and automate R&D workflows.

2020 — 2022

UST

Software Developer

Developed complex web portals for IoT-based connected vehicle projects. Pioneered the internal ReactJS Starter Kit used across multiple teams.

2016 — 2020

ITER, BBSR

B.Tech Computer Science

Graduated with a focus on Computer Science, mastering the fundamentals of data structures, algorithms, and distributed web systems.

Verified Endorsements

Endorsed by Engineering Leaders

Verified recommendations from engineering directors and tech leads at Dassault Systèmes.

LinkedIn Recommendation

"I managed Harsh during our time at Dassault Systèmes, where he served as a Front End Engineer with a clear interest in the evolving AI landscape. What sets Harsh apart is his practical knowledge of AI tools and agent orchestration, which he consistently applied to his front-end tasks. Harsh is a dependable developer who understands how to bridge the gap between standard web interfaces and AI-driven features. He would be a good addition to any team looking to incorporate AI agents into their technical stack."

NS
Nazarali Saliya
Director of Engineering at Dassault Systèmes
Managed Kumar directly
LinkedIn Recommendation

"As a Tech Lead, I worked closely with Harsh, who served as a UI Developer on our team. He demonstrated strong proficiency in React, modern JavaScript, and component-driven architecture, consistently delivering high-quality, maintainable UI solutions. Harsh has a solid understanding of UI/UX principles and translates designs into responsive, accessible, and performant interfaces. He is thoughtful about state management, reusability, and code standards, and he actively contributed to improving our frontend practices."

PL
Pradeep Londhe
R&D Engineer / Tech Lead at Dassault Systèmes
Managed Kumar directly
View Recommendations on LinkedIn ↗ linkedin.com/in/kumarharshn
Available for Select Contracts & Advisory

Let's build something exceptional

Have an AI agent, RAG pipeline, or full-stack MVP to ship? Let's talk scope, architecture, and timeline.

hello@kumarharsh.in ✉️