Senior Consultant · Enterprise R&D IT · Chennai, India

Vijayraj Palanisamy

C#/.NET engineer building GenAI and Agentic AI products

> C#/.NET Engineer. Roles: C#/.NET Engineer, Full-Stack Developer, GenAI & RAG, Agentic AI Builder, AI Product Builder.

I have spent more than 15 years delivering enterprise software — most recently R&D digital applications for a global life-sciences company. Alongside that work I design and ship my own AI products end to end: a PromptOps platform, an AI crypto intelligence site and app, an AI learning platform for YouTube, and privacy-first mobile apps.

View projects GitHub Contact

how my systems fit together

years in software, since 2011
15+years in software, since 2011
live AI web products I built
3live AI web products I built
Android apps on Google Play
2Android apps on Google Play
delivered in enterprise R&D IT
Bedrock · RAGdelivered in enterprise R&D IT

// about

From enterprise .NET to AI-native products

I started in 2011 implementing and integrating SAP Business One for clients, then spent three years as a .NET developer at GAVS Technologies building applications to client requirements.

Since 2016 I have worked for a global life-sciences company, first improving operational and digital processes, and since 2023 as a Senior Consultant in R&D IT / Digital DevOps. There I support R&D digital applications, modernize .NET applications, move workloads from on-premises to the cloud, build GitHub Actions pipelines and AWS Lambda monitoring, and delivered AWS Bedrock RAG question answering over vendor documentation.

I deliver work through specifications first. Clear requirements and acceptance criteria are the reference for everything after them, including the AI-assisted parts of the lifecycle, and engineers review every AI suggestion.

Outside work I build and run my own AI products end to end: architecture, back end, front end, mobile, infrastructure, CI/CD and operations. They range from a PromptOps platform with an MCP server to fully offline Android apps that run AI models on the phone. They are where I test, in production, the AI engineering practices I bring to a team.

// skills

Skills, by evidence — not by star ratings

No made-up proficiency percentages. Each skill shows where it was used: in employment, in my own products, or as current learning. Select a skill to see where.

Backend engineering

C#/.NET services and APIs, from client applications to clean-architecture SaaS back ends.

Frontend & mobile

Typed React front ends and Flutter apps with native Kotlin modules where performance or privacy needs it.

Data

Relational stores first, vectors where retrieval needs them, encryption where privacy demands it.

Cloud & DevOps

Automated delivery pipelines and infrastructure as code, with staged environments and rollbacks.

AI & machine learning

Retrieval, multi-model orchestration, agent tooling and on-device models — with deterministic fallbacks.

Architecture & security

Designs that keep the core logic in one place and make the risky paths explicit.

Delivery practice

Specifications first, AI-assisted where it helps, verified by tests before release.

Professional — Used in professional employmentHands-on projects — Used to build my own productsCurrently learning — Studying or experimenting

// experience

Career timeline

Fifteen years across ERP consulting, .NET development and enterprise R&D IT. Details of current employer work are kept confidential.

  1. Life sciences · R&D IT / Digital DevOps

    Digital engineering for R&D IT, working with cross-functional stakeholders on R&D digital applications, solution reliability and user experience.

    • Applied specification-driven development: clarified requirements, defined implementation-ready specifications and acceptance criteria, and aligned development and validation to the agreed scope.
    • Used an AI-assisted development lifecycle for requirements analysis, design decisions, implementation refinement, test-case preparation and developer-led review.
    • Implemented GitHub Actions CI/CD workflows and supported .NET application modernization and on-premises-to-cloud migration.
    • Used AWS Lambda for application monitoring with Microsoft Teams notifications.
    • Delivered AWS Bedrock and RAG-based question answering over vendor documentation.
    • C#
    • .NET
    • AWS Bedrock
    • RAG
    • AWS Lambda
    • GitHub Actions
    • Python
    • Microsoft Copilot Studio

// products & projects

AI and software products I have built

Real products with honest status labels. Live means publicly available now. Roadmap items are listed separately in each case study and are never presented as shipped.

tech:

8 projects shown

PromptFlowEngine homepage with the headline "Engineer better prompts. Connect AI to your codebase."★ Featured
Status: LiveAug 2026 – present

PromptFlowEngine

Prompt engineering, security and PromptOps platform for developers and teams

One deterministic TypeScript engine that analyses, secures and optimises prompts — shipped as a web workspace, HTTP API, Chrome and VS Code extensions, a CLI and an MCP server.

  • TypeScript (pure, one dependency)
  • js-tiktoken
  • Astro
  • React islands
  • Cloudflare Workers + Static Assets
  • Cloudflare D1
Case study Live site(opens in new tab)
CryptoTalkies homepage with the headline "Where crypto data becomes intelligence."★ Featured
Status: Live2025 – present

CryptoTalkies

AI-powered crypto news, education and market-intelligence platform — web and mobile

A free crypto intelligence site and Flutter app: live market data, AI signals, sentiment and whale tracking, a multi-provider AI agent, and automated news and education content.

  • Next.js (App Router, static export)
  • React
  • TypeScript
  • Tailwind CSS
  • Chart.js
  • Lightweight Charts
Case study Live site(opens in new tab)
AIYouTubeManager homepage with the headline "Learn anything. From YouTube. With AI."★ Featured
Status: LiveJun 2026 – present

AIYouTubeManager

Learn anything from YouTube, with AI — roadmaps, notes, quizzes and progress

An AI learning layer on top of YouTube: it builds a staged learning roadmap from a goal, matches videos to each stage, and writes notes, quizzes and flashcards from transcripts.

  • .NET 9
  • ASP.NET Core
  • Clean Architecture
  • CQRS / MediatR
  • EF Core
  • FluentValidation
Case study Live site(opens in new tab)
MSEW Infra ProjectsNode.js 24 · PostgreSQL · React
Status: Active development2026 – present

MSEW Infra Projects

Construction operations and RE wall production platform — web portal and Android app

Digitises site operations for reinforced-earth (RE) wall construction: geofenced attendance, daily progress reports, production and dispatch, materials, equipment and approvals.

  • Node.js 24
  • TypeScript
  • Express 5
  • Scheduler
  • Migrations
  • PostgreSQL
Case study Live site(opens in new tab)
AI PDF ScannerFlutter / Dart · CameraX · SQLCipher
Status: On Google PlayNov 2026 – present

AI PDF Scanner

Private, fully offline document scanner with OCR in English, Tamil and Hindi

An Android scanner that does everything on the device — capture, enhancement, OCR, searchable PDFs, encrypted storage and offline voice editing. The release build has no internet permission.

  • Flutter / Dart
  • Kotlin native modules
  • Platform channels
  • CameraX
  • OpenCV
  • Tesseract (English, Tamil, Hindi)
Case study Google Play(opens in new tab)
Land Price IntelligencePython · React · Flutter (Android, iOS)
Status: Working demoDec 2025 – present

Land Price Intelligence

Land research for buyers and investors — comparable sales, guideline values and infrastructure context

For any plot, shows seller asking prices, actual nearby sale prices and government guideline values separately, plus an evidence-based estimate range that refuses to guess when data is thin.

  • Python
  • FastAPI
  • SQLModel
  • SQLite (default)
  • Stripe (demo billing mode)
  • React
Case study
AI Sales HubASP.NET Core 10 (minimal APIs) · React 19 · xUnit
Status: Early developmentNov 2026 – present

AI Sales Hub

Multi-product AI sales and marketing agents with human approval on every external action

AI agents research, analyse, update an internal CRM and draft outreach. Nothing leaves the system without human approval and an evidence-based readiness gate.

  • ASP.NET Core 10 (minimal APIs)
  • ASP.NET Identity + MFA
  • EF Core + Npgsql
  • PostgreSQL 18
  • React 19
  • TypeScript
Case study
Offline AI Voice TranslatorFlutter · sherpa-onnx (VAD, ASR) · SQLCipher
Status: In developmentNov 2026 – present

Offline AI Voice Translator

Two-way speech translation that runs entirely on the phone

English, Hindi and Tamil speech recognition, translation and speech output in airplane mode, with no internet permission in the strict release build.

  • Flutter
  • Riverpod
  • go_router
  • Material 3
  • sherpa-onnx (VAD, ASR)
  • ONNX Runtime (Marian MT)
Case study

// architecture explorer

Explore how each product is built

Pick a product, then select any component to see what it does. Only implemented components are shown. Secrets, private endpoints and internal network details are left out.

One Cloudflare Worker serves the static Astro site from the edge and runs only for /api/* requests. Every client embeds the same engine package.

Clients
Edge
Core
Data
AI
Delivery

// how i build software

Specification first. Automated all the way to production.

The same lifecycle I use at work and on my own products, plus where AI fits into it — with practised steps clearly separated from experimental ones.

Delivery lifecycle

Established practice · 15+ years

Understand — Business needs

  • Stakeholder discovery
  • Business requirements
  • Constraints and risks

The loop repeats: what I learn in Operate feeds the next Understand.

AI-native lifecycle: how AI fits into the work

● Practised: at work and in my products◌ Experimental: my own products only

AI speeds up each stage, but the specification and the people stay in charge. The AI column describes assistance, not autonomy, except in the last row, which is still experimental.

StepEngineerAI assistanceStatus
SpecificationOwns the intent, scope and acceptance criteriaFinds ambiguity and missing edge cases● Practised
Context & promptChooses the relevant code and constraintsBuilds a structured, token-efficient prompt from the spec and codebase● Practised
Design & implementDecides the architectureProposes designs and drafts code against the spec● Practised
Review & testReviews every change against the specPrepares test cases and review checklists● Practised
GuardrailsSets the policyScans prompts and agent tool calls for secrets, PII and injection● Practised
Agent executionApproves every external actionAgents research, draft and update internal systems◌ Experimental

architecture case study · PromptFlowEngine

One engine, edge-hosted, no prompt storage

  • Context: the same prompt rules are needed in a browser, editor, CLI, agents and an API.
  • Decision: one pure-TypeScript engine package; one Cloudflare Worker serves the static site and runs only for /api/*.
  • Rejected: Pages + Functions (a second routing model) and a separate backend (the product has no server-side prompt state).
  • Result: per-PR previews, a gated production deploy, instant rollback, and server-side metadata only.
Full case study →

architecture case study · AIYouTubeManager

Clean-architecture .NET with quota-aware ingestion

  • Context: YouTube Data API quota is too small to poll every channel for every user.
  • Decision: a background worker syncs new videos from RSS feeds at zero quota; a budget service rations Data API calls.
  • Retrieval: pgvector embeddings in the same PostgreSQL database keep RAG simple to operate.
  • Delivery: Terraform for AWS, with GitHub Actions for CI, deploys, Lighthouse and a dedicated rollback workflow.
Full case study →

// ai engineering journey

From enterprise RAG to agents and on-device AI

How my AI work has grown, step by step. Professional delivery, my own products and ongoing learning are labelled separately.

  • Generative AI & LLMs
  • RAG & semantic retrieval
  • Prompt engineering & PromptOps
  • Model Context Protocol
  • AI agents with human approval
  • AI security & guardrails
  • AI-assisted programming
  • Specification-driven development
  • On-device AI
  1. 2011 – 2015Professional

    Foundations: ERP and .NET

    Started with SAP Business One implementations and integrations, then moved into .NET application development for client projects.

    • C#
    • .NET
    • System integration
  2. 2016 – 2023Professional

    Life-sciences operations and digital process improvement

    Workflow analysis, data-driven reporting and process improvement across cross-functional teams.

    • Process improvement
    • Reporting
  3. 2023 →Professional

    Generative AI in the enterprise

    Delivered AWS Bedrock RAG question answering over vendor documents, used Microsoft Copilot Studio, and applied an AI-assisted, specification-driven delivery lifecycle in enterprise R&D IT.

    • AWS Bedrock
    • RAG
    • Copilot Studio
    • SDD
  4. 2025 →My products

    Multi-provider AI in a consumer product

    CryptoTalkies: an AI agent (ARIA) that combines live tools with Claude, Gemini, OpenAI and Groq and falls back between providers, plus automated content pipelines.

    • LLM orchestration
    • Tool use
    • Automation
  5. Jun 2026 →My products

    RAG on my own stack

    AIYouTubeManager: pgvector embeddings, semantic search and an AI mentor chat grounded in the user’s own channels, on a clean-architecture .NET API.

    • RAG
    • pgvector
    • Embeddings
    • .NET
  6. Aug 2026 →My products

    PromptOps, MCP and agent guardrails

    PromptFlowEngine: a deterministic prompt engine, an MCP server with 19 tools, agent hooks and an MCP tool-call firewall, and CI gating for prompts.

    • Prompt engineering
    • MCP
    • AI security
    • PromptOps
  7. Nov 2026 →My products

    On-device AI and human-in-the-loop agents

    Offline OCR and speech translation on Android, and an agentic sales system where every external action needs human approval.

    • On-device AI
    • ONNX
    • Agentic AI
    • Human approval
  8. In progressLearning

    Agentic AI & Applications — IIT Madras

    Formal study of agentic AI architectures and applications. In progress; not a completed certification.

    • Agentic AI
    • Multi-agent systems

// contact

Let’s build something reliable — with AI where it helps

I’m interested in

  • Senior software engineering roles (C#/.NET, full stack)
  • AI engineer and Agentic AI engineer roles
  • Technical lead and architecture discussions
  • Product collaborations and consulting
Email me