01Learn
Foundations in data and software
Started by building with data, Python, and software fundamentals — learning to reason about structure before reaching for tools.
Sanidhya RathoreAI ENGINEER · BUILDER
Building
intelligence
into products.
I build AI-powered products and full-stack systems from interface to infrastructure.
Currently buildingintelligent financial infrastructure
- SystemOnline
- AIActive
- StackFull-stack
Scroll to explore
I don’t just
build interfaces.
I build systems.
I work across the stack — from React and Next.js interfaces to Python APIs, databases, AI workflows and deployment. That allows me to move from an idea to a working product without losing the connection between product thinking and engineering.
Working across the whole stack means the interface and the data model stay in the same conversation. It removes the handoff where design decisions quietly become engineering problems.
System Architecture
From the user
down to the metal.
Every product is the same six layers. I work across all of them, because the interesting problems live in the seams between them.
- UserThe person and the constraint
- InterfaceWhat they see and interact with
- ApplicationOrchestration and state
- AI / LogicReasoning and rules
- DataWhat is actually true
- InfrastructureWhere it runs and how it ships
Built by
building.
Not a résumé — a progression. Each phase was a deliberate move from understanding a layer to owning what sits around it.
- 02Ship
Real-world software development
Moved into professional development and full-stack product engineering, working inside production codebases with real constraints.
- 03Build Systems
Across the entire stack
Worked through frontend, backend, databases, authentication, APIs, and deployment — owning the seams between them rather than one layer.
- 04Explore AI
Intelligence as a component
Started building AI-powered systems, automation workflows, and intelligent products where the model is a part of the system, not the whole system.
- 05Build for Real Users
AI × automation × finance
Current direction: intelligent financial infrastructure — systems that understand financial information, evaluate risk, and keep a human in the loop.
What I build.
Four areas where the work overlaps — which is the point. A system that spans more than one of these is where the real engineering lives.
AI Systems
Retrieval pipelines, multi-agent orchestration, and structured outputs that stay grounded in evidence.
- LLMs
- Agents
- RAG
- Automation
- Structured Outputs
- Tool Calling
Product Engineering
Typed, accessible interfaces and the APIs behind them — built to be maintained, not just shipped.
- Next.js
- React
- TypeScript
- Python
- FastAPI
- REST APIs
Data & Infrastructure
Schemas, migrations, auth, and deployment — the parts that decide whether a system survives contact with reality.
- PostgreSQL
- MySQL
- SQL
- Docker
- Authentication
- RBAC
Intelligent Products
Applied AI where the decision matters more than the output: risk, evidence, and human approval.
- AI Finance
- Computer Vision
- Decision Systems
- Workflow Automation
- Developer Tools
Selected
work.
Case studies rather than cards — the problem, the approach, and what the system actually does.
PROJECT 01AI Systems · Retrieval · Decision Support · 2026 · DEPLOYED
Aptino
An evidence-grounded claim decision engine for health insurance.
Tech Stack
- Python
- FastAPI
- Streamlit
- Hybrid Retrieval
- BM25
- RRF
- Reranking
- Docker
- LLM Agents
Problem
Claim adjudication depends on reading a large policy document and matching each fact in a claim to the clause that governs it. Doing that by hand is slow, and a summary that loses its citations cannot be defended in an audit.
Approach
Treat the system as investigation rather than prediction. Agents normalise the claim, plan what evidence is missing, retrieve the governing clauses, build an explicit evidence matrix, then reason over it under a fail-closed rule: when evidence is missing or conflicting, the system abstains and returns NEEDS_REVIEW instead of guessing.
Architecture
- Case Analysis Agent normalises facts and flags conflicts and missing evidence
- Hybrid retrieval: dense vectors and BM25 fused with Reciprocal Rank Fusion
- Post-fusion reranker narrows to high-precision policy chunks
- Policy Evidence Agent builds the evidence matrix with typed findings
- Decision Agent applies a safety hierarchy: Missing → Exclusion → Limits → Admissible
- Validation Agent checks every material claim against a retrieved citation
- FastAPI endpoints (/analyze, /review, /health) behind a Streamlit reviewer workspace
- Full provenance chain recorded as audit events for every analysis
More Projects
- 022026 · DEPLOYED
Vishwakarma Arts
A deployed React web application.
- JavaScript
- React
- Vite
- Vercel
- 032026 · DEPLOYED
Portfolio
This site — an interactive engineering demonstration.
- Next.js
- React
- TypeScript
- Three.js
- 042026 · EXPERIMENT
iTunes Music Sales Analysis
A data analysis study over a public sales dataset.
- Python
- Data Analysis
The stack.
Tools I reach for, and what each one is for. Hover a node to see it in context.
Languages
Frontend
Backend
AI
Data
Infrastructure
The way
I build.
Three principles that decide what gets built, and what gets thrown away before it wastes a month.
Start with the problem
Technology comes after understanding the user and the constraint. Choosing the stack first is how you end up building the wrong thing well.
Ship the smallest real system
Build something usable before building something impressive. A running end-to-end slice teaches more than an elaborate plan.
Make the system explainable
AI should produce evidence, structure, and observable decisions — not just an output. If nobody can see why it decided, it isn't finished.
Building
in public.
Pulled directly from the public GitHub record — no screenshots, no claims.
What I’m
building toward.
I'm interested in the intersection of AI, automation and financial infrastructure — systems that can understand financial information, evaluate risk, and automate decisions while keeping humans in control.
EXPLORING
AI × FINANCE × AUTOMATION
This is a direction, not a product with customers. The point of the pipeline is to make the intended shape of the system explicit — where automation ends, and where a person still decides.
Intended Pipeline
- Financial Data
- AI Analysis
- Risk Signal
- Decision
- Human Approval
- Action
Experience and current status
Experience
2025
Software Development
Sthapatya Consultants · Government / civic technology
Built and maintained modules across a civic technology codebase — frontend engineering, database interaction, access control, and debugging production workflows against real operational requirements.
- Built and extended frontend modules in React and TypeScript
- Worked directly against SQL databases and schema constraints
- Implemented and maintained authentication and access control
- Debugged production workflows and resolved defects in live systems
- Coordinated interface and backend changes across module boundaries
- Next.js
- React
- TypeScript
- SQL
Currently
- Building
- AI × Finance × Automation
- Learning
- AI SystemsDistributed SystemsProduct Engineering
- Open to
- FoundersBuildersInteresting problemsAI product opportunities
About
Sanidhya Rathore
AI Engineer · Full-Stack Builder · Product Engineer
Pune, India
I enjoy taking ambiguous problems and turning them into working software.
Contact
Have a problem
worth building?
The most useful thing you can send is a description of the problem you are actually stuck on.