Skip to content

ShyamKumar1/ShyamKumar1

Repository files navigation

Typing SVG



AI Engineer & Senior Python Developer with 4+ years of experience spanning enterprise software development at Cognizant, AI model evaluation at Labelbox (Alignerr), CTO leadership at Technitude Info Solutions (managed 15+ engineers, delivered 50+ client projects), and entrepreneurial brand leadership as Founder of Trilimedia Digital Agency.


What I Engineer

  • AI / ML Systems — LLM evaluation, prompt engineering, RAG pipelines, model benchmarking
  • Full Stack Platforms — Cloud-native microservices, enterprise web apps, real-time data systems
  • Product Engineering — From zero-to-one MVPs to enterprise platforms with 99.9% uptime

Open To





Languages

Frontend

Backend & Databases

Cloud, DevOps & Tooling



Domain Proficiency Details
LLM Evaluation Advanced Model output scoring, safety benchmarking, edge-case identification, evaluation scenario design at $80-100/hr
Prompt Engineering Advanced Jailbreak testing, system prompt optimization, multi-turn attack vectors, structured output design
RAG & Retrieval Intermediate Vector databases, document chunking strategies, hybrid search, embedding optimization
AI Model Serving Intermediate Model deployment pipelines, API wrappers, latency optimization, A/B testing frameworks
Data Engineering Intermediate ETL pipelines, data processing automation, feature stores, n8n workflow orchestration
Enterprise Full Stack Advanced Spring Boot + Angular enterprise apps, cloud-native microservices, CI/CD, 80%+ test coverage


Alexander — Autonomous AI Agent Framework

A production-grade autonomous AI agent framework with three-tier memory architecture (working, episodic, semantic), real-time voice pipeline via VAD+Whisper, and a sophisticated goals scheduler with priority queues. Built entirely from scratch to bridge AI research and deployable engineering.

Attribute Detail
Stack Python, FastAPI, WebSocket, Redis, PostgreSQL, Docker
Scale Multi-agent orchestration with async event-driven communication
Performance Sub-100ms inference pipeline; VAD-triggered wake word detection
Security Sandboxed execution, input validation, rate-limited endpoints
Impact Full-featured AI assistant framework ready for production deployment
Repository github.com/ShyamKumar1/alexander

Three-tier memory architecture, goals scheduler with priority queues, real-time voice interface via VAD+Whisper pipeline, and modular plugin architecture.

AI Code Evaluator — Automated Code Quality Scoring

An automated framework that scores AI-generated code across four weighted dimensions — correctness (40%), edge cases (25%), performance (20%), and code quality (15%) — producing a 0-100 score with detailed Markdown reports. Inspired by Labelbox and Scale AI evaluation pipelines.

Attribute Detail
Stack Python 3.11+, pytest, PyYAML, GitHub Actions CI
Scale Multi-dimensional scoring matrix with configurable weight profiles
Performance Batch evaluation pipeline with parallel test execution
Security Sandboxed code execution, input validation, CI-scoped permissions
Impact Production evaluation tool used for AI model benchmarking workflows
Repository github.com/ShyamKumar1/ai-code-evaluator

100% pytest coverage with CI matrix across Python 3.11-3.13. Configurable rubric weights, detailed scoring breakdowns, and pass/fail thresholds for automated quality gates.

DSS Accounting System — GST-Compliant Silk Industry Platform

A full-stack monorepo accounting system purpose-built for the silk saree industry. Features double-entry accounting, GST engine (CGST/SGST/IGST), invoice management, credit notes, bank reconciliation, and GSTR-1/3B reporting for 5 firms. Turbo monorepo with Prisma ORM and PostgreSQL.

Attribute Detail
Stack TypeScript, Turbo, Next.js, Fastify, Drizzle ORM, Turso, Prisma, PostgreSQL, Redis, Docker
Scale Multi-tenant architecture serving 5 firms with firm-level data isolation
Performance Redis-cached queries; sub-100ms report generation for GSTR filings
Security Row-level security, session-based auth, audit logging for all financial transactions
Impact Production system managing real financial data for multiple businesses
Repository github.com/ShyamKumar1/dss

End-to-end GST compliance engine: auto-calculates CGST/SGST/IGST based on transaction type. Generates GSTR-1 and GSTR-3B returns. Multi-firm dashboard with role-based access control.

Money Machine — Revenue Automation & Lead Engine

A revenue automation platform with Express dashboard, n8n workflow engine, and three active automation pipelines — Lead Gen ($197-497/mo), Content Engine ($147-247/mo), and Booking System ($97-197/mo). Processes 1,000+ qualified leads with real-time dashboards.

Attribute Detail
Stack Python, Express.js, n8n, SQLite, REST APIs
Scale 1,000+ leads processed with automated enrichment and scoring
Performance Real-time lead ingestion with sub-minute processing latency
Security API-key authenticated webhooks, rate-limited endpoints, encrypted storage
Impact Revenue-generating system with active client subscriptions
Repository github.com/ShyamKumar1/money

Three-tier automation pipeline: lead sourcing & enrichment → content generation & distribution → booking & follow-up automation. n8n orchestrates the entire workflow with Hermes handling data processing.

Trilimedia Brand Projects — Tropica Café & TARA Jewelry

Flagship brand identity and content strategy projects delivered through Trilimedia Digital Agency. Tropica Café was repositioned as a premium café brand with full visual identity and 6-month content system. TARA Jewelry received a complete brand identity suite with AI-augmented documentation and structured launch architecture.

Brand Deliverables
Tropica Café Brand repositioning, visual identity system, 6-month content strategy, social media framework, story-led engagement model
TARA Jewelry Full brand identity suite, logo system, design templates, storytelling assets, structured launch campaign architecture

Both projects delivered through Trilimedia Digital Agency (founded 2023). Combined result: measurable customer engagement lift and zero off-brand incidents across all accounts.

Hermes Vault — Automated Knowledge Management

An automated knowledge management and archiving platform that captures, organizes, and surfaces daily engineering insights, project artifacts, and research findings through an elegant web interface. Runs on a nightly cron-driven build pipeline.

Attribute Detail
Stack TypeScript, Next.js, Tailwind CSS, Markdown, Netlify, GitHub Actions
Scale Automated daily ingestion with cron-driven builds
Performance Static site generation; sub-second page loads globally via CDN
Security API-gated write operations, HTTPS everywhere
Impact Centralized knowledge base serving as daily engineering journal
Repository github.com/ShyamKumar1/hermes-vault

End-to-end automated pipeline: GitHub Actions cron triggers content ingestion → Markdown generation → static site build → Netlify deployment. Sections for daily logs, blogs, projects, contracts, and archive.


Senior Python Developer — AI Model Review

Jun 2026 — Present · Remote

  • Evaluating generative AI models on Alignerr platform — reviewing LLM outputs for correctness, safety, and quality
  • Designing evaluation scenarios to identify model failure modes across diverse domains at $80-100/hr
  • Scoring model performance across accuracy, edge-case handling, and code quality dimensions


Chief Technology Officer (CTO)

Aug 2025 — Jan 2026 · India

  • Defined technology roadmap; led 15+ engineers across Python, DevOps, QA, and delivery teams
  • Architected AI-powered data pipelines reducing client manual workflows by up to 60%
  • Established pytest (80%+ coverage), GitHub Actions CI/CD, mandatory code review standards
  • Delivered 50+ client projects with 99.9% uptime on AWS/GCP (Docker, Kubernetes, Terraform)


Founder & CEO

2023 — Present · India

  • Founded full-service digital agency serving 10+ clients across F&B, retail, IT, fitness, and education sectors
  • Delivered brand identity systems, content governance frameworks, statewide campaigns, and creative strategy
  • Built content workflows and brand frameworks achieving zero off-brand incidents across all accounts


Programming Analyst

Mar 2022 — Nov 2024 · India

  • Developed enterprise web applications using Java (Spring Boot), Angular, TypeScript, and AWS
  • Collaborated in large multi-team Agile environments (JIRA, GitHub) delivering to production quarterly
  • Star of the Quarter Award for exceptional contribution and delivery quality



Recognition Details
Cognizant award for exceptional contribution, delivery quality, and team collaboration
Delivered as CTO at Technitude — maintained 99.9% uptime across all deployments
Contributions to Marimo, LM Studio Bridge, OpenCode, and Hermes Agent ecosystem
Active Labelbox/Alignerr contractor evaluating frontier LLMs at $80-100/hr
Founded Trilimedia Digital Agency — 10+ clients, zero off-brand incidents


AWS

IBM

Google Cloud











GitHub Contribution Snake


learning:
  - Advanced LLM evaluation methodologies
  - Distributed systems design patterns
  - AI safety and alignment research

building:
  - Autonomous AI agent frameworks (Alexander)
  - Revenue automation pipelines (n8n + Hermes)
  - Production-grade AI evaluation tooling

exploring:
  - AI red teaming and bug bounty programs
  - Multi-modal AI architectures
  - Brand strategy + AI-augmented content systems

open_to:
  - Senior AI Engineer / Python Developer roles
  - Technical consulting and fractional CTO engagements
  - Open source collaborations
  - Speaking and community contributions




"Engineering is not just about writing code — it's about architecting systems that make a difference."


About

Profile intro

Topics

Resources

Stars

0 stars

Watchers

1 watching

Forks

Contributors