Jona Joy
I build for production, own what I ship, and make real value flow.
Backend & Systems Engineer · AI/ML & Agent Systems · Blockchain & Tokenization
3
Years Building
50%
Throughput Gain
40%
Faster CI/CD
0→1
Founding Engineer
01 / ABOUT
Founding engineer.
Still building.
Jona Joy
Kochi, Kerala, India
I'm a backend engineer who builds systems from zero to production — high-throughput, secure, and designed to scale under real load. As one of the first three engineers at Kireap, I've spent three years in the full weight of early-stage ownership: architecture decisions, production incidents, business trade-offs, customer conversations, and shipping things that actually work.
My core is backend infrastructure — FastAPI, Golang, PostgreSQL, Redis, AWS, event-driven systems. I currently lead three technical tracks simultaneously: proprietary algorithm development (hands-on, not just oversight), payments and PSP integrations (gateways, payouts, settlements, custom features like name-matching), and blockchain tokenization initiatives. I've also built full-stack (Next.js, React, Flutter) and led ML work involving transformers and computer vision when the problem called for it.
I think and execute like an owner. I'm deeply drawn to AI/ML systems and agent development — that's where I'm building next.
PROFILE
EDUCATION
GPA 9.33 / 10B.Tech, Information Technology
School of Engineering, CUSAT
Class of 2024
WHAT DRIVES ME
- →Systems that hold under real pressure, not just greenfield benchmarks
- →Ownership over assignment — I build like it's my product, not my ticket
- →AI that solves real problems in production, not just impresses in demos
- →Startup instincts: field execution, user empathy, and business trade-offs alongside the code
FOUNDER STORY
0 → 100One of 3 engineers · Kireap · 2023 – present
In 2023, as a 3rd-year engineering student, I joined Kireap as one of three engineers with one clear goal: build something real, from zero. We didn't just write code — we did product thinking, field demos, customer conversations, and operational execution with a team of 3–4 people having founder-level discussions every day. Between October–November 2023, we traveled to Madhya Pradesh, visited farms, spoke directly with farmers about access, pricing, and trust — and built with what we learned. We launched an MVP in December 2023. Then we scaled: 0→1, 1→10, 10→100 — across payments, blockchain, AI/ML, and Industry 4.0. That journey gave me more than technical depth. It gave me startup resilience, product intuition, and the ability to own outcomes — not just tasks.
Full journey on LinkedIn →02 / FEATURED WORK
Projects that
ship and scale.
AI Status Agent
Agentic RAG system for engineering project intelligence
Enterprise-grade agentic AI backend that answers project status queries by reasoning over tickets, Slack messages, and docs. Uses a LangGraph ReAct-style agent with multi-tool selection — not just an LLM wrapper, but a full reasoning system with query rewriting, conversation memory, and an observability trace endpoint.
- ·LangGraph ReAct agent with dynamic tool selection across tickets, Slack, and docs
- ·Full RAG pipeline: ChromaDB vector store + Ollama local inference, zero external API calls
- ·Clean architecture (API → Application → Domain → Infrastructure) with /trace observability endpoint
Satellite Change Detection
Government hackathon: detecting human-made changes from orbit
Built for the DPIIT Hackathon organized by IIT Delhi — a full production pipeline for detecting human-made changes in satellite imagery using time-series data. Not just R&D: we shipped the complete system including ML architecture, backend API, image preprocessing, and annotation pipelines.
- ·Siamese Twin Transformer architecture for bi-temporal satellite image pair comparison
- ·FastAPI backend pipeline with image preprocessing, annotation management, and inference serving
- ·Detects construction, roads, demolition — distinguishing human-made changes from natural variation across time-series imagery stacks
PRISM Finance
On-chain lending without over-collateralization
A decentralized lending protocol that replaces over-collateral requirements with on-chain reputation scoring and DAO-backed collateral pools — making DeFi credit accessible. Designed the full smart contract architecture and wired it to a production-grade Next.js frontend.
- ·Custom reputation-based credit scoring engine in Solidity
- ·DAO-backed collateral model with automated loan issuance
- ·Live on Sepolia testnet with full-stack dApp on Vercel
InvoChain
Invoice tokenization for SME liquidity
A Web3 invoice factoring platform enabling small businesses to tokenize receivables and access immediate liquidity through investor marketplaces. Handled smart contracts, backend services, and the full transaction workflow end-to-end.
- ·ERC-compatible invoice token contracts with automated settlement
- ·FastAPI + PostgreSQL backend for transaction and user management
- ·Secure buy/sell flows with integrated payment distribution logic
clariFi
Personal CFO powered by verifiable AI
A financial decision engine that acts as a personal CFO — combining verified, deterministic calculations with AI-assisted explanations. Built to give users confidence in every financial decision by making the math transparent and the reasoning auditable.
- ·Verifiable calculation engine: all financial outputs are deterministic and explainable, not LLM guesses
- ·AI layer explains complex decisions in plain language without hallucinating figures
- ·Personal CFO UX — budgeting, forecasting, and decision support in a single interface
MedVoice Agent
Production voice AI for hospital workflows
A production-quality hospital voice assistant with structured intent routing — patients speak naturally to book appointments, look up records, or check availability. Fully pipelined from Whisper STT through Ollama LLM reasoning to TTS output, with a React frontend and strict hallucination-prevention guardrails.
- ·Whisper STT with silence detection + structured intent routing (appointments, availability, patient lookup)
- ·Strict LLM hallucination prevention — only uses verified mock-DB data, never invents patient records
- ·Full-stack: FastAPI async backend + React/Tailwind frontend with real-time voice controls
FitMate
Pose tracking and calorie estimation with real-time ML
A fitness tracking app that uses ML Kit pose detection to track body movement in real time — automatically counting reps, estimating calories burned, and logging workouts without manual input. Flutter frontend backed by FastAPI and PostgreSQL.
- ·ML Kit pose estimation for real-time body landmark tracking and rep counting
- ·Calorie estimation engine using pose data, movement velocity, and exercise classification
- ·Flutter frontend with FastAPI backend and PostgreSQL for session history and progress tracking
03 / EXPERIENCE
Where I've
built things.
Software Engineer — Core Technical Team
Kireap Technologies
- →Technically lead the proprietary algorithm team — hands-on architecture, system design, and performance optimization for core pricing and algorithmic components used across the system.
- →Lead payments and PSP integration end-to-end: gateway connections, payout flows, settlement logic, and custom-built features including name-matching and reconciliation — owning both the technical build and direct business coordination with external PSPs.
- →Built the full traceability stack hands-on in Polygon, then led its migration to Hyperledger Fabric — increasing transaction throughput by 50% with enforced data privacy through a permissioned architecture.
- →Led the tokenization initiative at Kireap (financial and asset tokenization) — defined the full product journey, flows, and integrations; managed and guided the development team through execution as the technical product owner.
- →Redesigned microservices architecture to double transaction throughput under production load while maintaining availability SLAs.
- →Hardened system security via JWT-based auth, AWS infrastructure guardrails, and end-to-end encryption across all sensitive data paths.
- →Automated CI/CD pipelines with Docker, reducing deployment time by 40% and eliminating a class of manual-error production incidents.
Software Engineer Intern
Kireap Technologies
- →Joined as one of three engineers at an early-stage startup — contributed across product thinking, architecture, field execution, and customer-facing demos alongside technical build.
- →Architected and delivered a full-stack system from scratch (0→1): sole engineer responsible from system design through MVP launch in December 2023.
- →Built production modules across Next.js, Flutter, FastAPI, and Solidity — enabling a modular, cross-platform product foundation across multiple user bases.
- →Traveled to Madhya Pradesh for field dry runs with farmers (Oct–Nov 2023), gathered direct user feedback, and iterated the product to launch.
- →Delivered payment and user-management prototypes that accelerated stakeholder validation and compressed feedback cycles.
04 / SKILLS
Built with the
right tools.
Languages
Backend & Infra
Frontend & Mobile
Blockchain
AI & Data
DevOps & Cloud
Always learning, always shipping.
Currently deepening: LLM agent architectures, ML pipelines, and system-level performance engineering.
05 / CONTACT
Let's build
something real.
Whether you're hiring, exploring a technical collaboration, or want to talk backend systems and AI — I'm up for it.
Open to opportunities
Particularly interested in backend-heavy roles at the intersection of AI systems, payments infrastructure, or agent development. Teams that ship with intention and move fast.
Send a message →