Welcome to my profile
Pramod Kumar Voola
AI Builder
Engineering Manager at Medidata Solutions (Dassault Systèmes)
“The best way to predict the future is to build it, with systems that learn, adapt, and scale.”
Quoting Bobby Holley
As part of our continued collaboration with Anthropic, we had the opportunity to apply an early version of Claude Mythos Preview to Firefox. This week’s release of Firefox 150 includes fixes for 271 vulnerabilities identified during this initial evaluation. [...] Our experience is a hopeful one for teams who shake off the vertigo and get to work. You may need to reprioritize everything else to bring relentless and single-minded focus to the task, but there is light at the end of the tunnel. We...
Changes to GitHub Copilot Individual plans
Changes to GitHub Copilot Individual plans On the same day as Claude Code's temporary will-they-won't-they $100/month kerfuffle (for the moment, they won't), here's the latest on GitHub Copilot pricing. Unlike Anthropic, GitHub put up an official announcement about their changes, which include tightening usage limits, pausing signups for individual plans (!), restricting Claude Opus 4.7 to the more expensive $39/month "Pro+" plan, and dropping the previous Opus models entirely. The key...
Is Claude Code going to cost $100/month? Probably not - it's all very confusing
Anthropic today quietly (as in silently, no announcement anywhere at all) updated their claude.com/pricing page (but not their Choosing a Claude plan page, which shows up first for me on Google) to add this tiny but significant detail (arrow is mine, and it's already reverted): The Internet Archive copy from yesterday shows a checkbox there. Claude Code used to be a feature of the $20/month Pro plan, but according to the new pricing page it is now exclusive to the $100/month or $200/month Max...
About
Pramod Kumar Voola
Pramod Kumar Voola is an Engineering Manager at Medidata Solutions (Dassault Systemes) who builds at the intersection of AI systems and engineering leadership. His work centers on agentic workflows, automation platforms, and cloud-native applications -- designed for production from day one.
He takes a first-principles approach to software: decompose hard problems, engineer for failure modes early, and scale deliberately from prototype to enterprise. The portfolio itself reflects this philosophy -- built with Claude Code agent teams on Next.js, TypeScript, Python, and AWS infrastructure.
Beyond the day job, Pramod maintains a Tech Radar tracking emerging tools, a Cookbook of 18 engineering patterns, and Neural Stream -- an AI-curated daily news feed with an accompanying newsletter. This site is both portfolio and working system.
Languages
Frontend
Backend
AI/ML
Cloud
Data
DevOps
Tools
Beyond the Code
Reading
Favorite Pastime
Always Exploring
Intelligence Hub
My personal knowledge base. The Tech Radar is where I publicly track my stance on the tools and frameworks I'm building in pipeline or already built, evaluate in sandboxes, or actively move away from. The Cookbook collects reusable engineering patterns for AI/ML challenges I solve repeatedly.
Industry Tech Radar
Where the industry is heading in 2026. Curated from ThoughtWorks Radar, Stack Overflow Survey, GitHub Octoverse, and leading engineering blogs. Adopt = battle-tested in production. Trial = actively experimenting. Assess = on the watchlist. Hold = phasing out or avoid for new work.
MCP (Model Context Protocol)
↑AI/ML & Agents
Curator Note: 📝 Building an MCP server for this site. The protocol is elegant and the ecosystem is growing fast. This is the USB-C of AI tooling.
Source: ThoughtWorks Vol 33 top theme
Context Engineering
↑AI/ML & Agents
Curator Note: 📝 Prompt engineering was level one. Context engineering is the real game. Controlling what the model sees matters more than how you phrase the question.
Source: ThoughtWorks Vol 33: replaced prompt engineering
Claude (Anthropic)
↑AI/ML & Agents
Curator Note: 📝 My default LLM for everything from code to analysis. The extended thinking capability is unmatched for complex reasoning tasks.
Source: Stack Overflow 2025: most admired LLM
RAG (Retrieval-Augmented Generation)
→AI/ML & Agents
Curator Note: 📝 RAG is table stakes for any production AI system. I prefer chunked retrieval with re-ranking over naive similarity search. The pattern is mature; the tuning is where the art lives.
Source: Mature pattern with established ecosystem
AI Code Assistants
↑AI/ML & Agents
Curator Note: 📝 AI-assisted coding is no longer optional. I use Claude Code as my primary tool and the agentic mode is a genuine force multiplier. Every engineer should be fluent in at least one.
Source: 80% GitHub users trying Copilot/Claude Code
LangGraph
↑AI/ML & Agents
Curator Note: 📝 Graph-based orchestration solves the control-flow problem that linear chains cannot. I use it for every workflow that needs loops or human-in-the-loop steps.
Source: Leading agent orchestration framework
Agentic Workflows
↑AI/ML & Agents
Curator Note: 📝 I run agent teams daily for parallel coding tasks. The productivity gain is real, but you need guardrails. Unreliable without structured outputs and checkpoints.
Source: Deloitte 2026 transformative force
LLM Observability (LangSmith/Langfuse)
↑AI/ML & Agents
Curator Note: 📝 You cannot improve what you cannot measure. I log every chain execution in LangSmith. The eval framework alone justifies the investment.
Source: Critical for production AI systems
Amazon Bedrock
↑AI/ML & Agents
Curator Note: 📝 Enterprise-ready model infrastructure. The knowledge base integration with S3 is particularly useful. Good for teams already invested in AWS.
Source: Managed multi-model LLM infrastructure
Multi-Agent Systems
↑AI/ML & Agents
Curator Note: 📝 Exciting but still early. Most multi-agent demos break down on real workloads. I am watching CrewAI and AutoGen closely but not betting production on them yet.
Source: AutoGen, CrewAI, Swarm: early but growing
Artifacts
portfolio.os
LiveAn intelligence hub and living portfolio with zero-cost hosting.
Architecture: Next.js 15 static export with EventBridge-triggered curation pipeline and AI summaries. MCP server exposes portfolio data as tools for AI clients.
Impact: Automated personal brand maintenance, dynamically generating a daily AI newsletter without manual writing.
Nebula Protocol
LiveCRDT-based consensus engine that eliminates phantom commitments in multi-agent AI systems via the Google A2A protocol.
Architecture: Agent-State CRDTs (monotone lattices, LWW registers, OR-Sets) with cryptographic observation proofs, causal barriers, and gossip-based delta-sync. FastAPI core with React dashboard. Reduces inter-agent token costs by ~97%.
Impact: Reduces phantom commits from 5.4% to 0.1% in multi-agent systems with sub-750ms p99 convergence and 100% availability during network partitions — no central coordinator needed.
Neural Stream Curator
LiveAutomated content curation pipeline that surfaces, summarizes, and tags AI news daily.
Architecture: Python pipeline with 20+ RSS feeds, LLM-powered AI summaries, and Resend for email delivery. Weekly deep-dive with tech radar auto-updates.
Impact: Processes 500+ entries daily into an opinionated digest, eliminating 2+ hours of manual industry research.
Agentic SDLC Framework
LiveAI agents embedded across planning, implementation, debugging, review, and release. IDE-agnostic. Runtime-backed. Evidence-first.
Architecture: Structured session runtime with plans, traces, approvals, and events. Debug fabric captures command, test, CI, and browser verification evidence in a common schema. Plugin packs for debugging, browser, security, compliance, JIRA, and semantic code tools.
Impact: Machine-enforced quality gates, bounded token architecture, config-driven workflows. 98% industry standard compliance (IEEE 12207, OWASP, NIST SSDF, DORA, ISO/IEC 25010, ALCOA+).
AgentVCR
LiveRecord and replay MCP client-server traffic for deterministic testing. No live servers, rate limits, or drift.
Architecture: Python + TypeScript dual SDK. Records MCP traffic into .vcr cassettes via event-sourcing, replays in tests and CI with full state snapshots. Shared cassette format across languages.
Impact: Eliminated flaky MCP integration tests. Built for compatibility gates, regression detection, and deterministic CI pipelines that mature platform teams expect.
Agentic Handbook
In ProgressAn evolving guide to building systems that think, plan, and act.
Architecture: MDX-powered documentation with interactive code examples, decision trees, and architecture diagrams.
Impact: Open-source knowledge base covering first principles through production deployment of AI agent systems.
GantryFleet
LiveThe Headless Software Factory. A CLI-first autonomous agent that turns voice/text commands into architected code, self-healing tests, and deployed PRs.
Architecture: Python-based CLI agent with voice input, code generation, automatic test writing with self-healing, and PR deployment. Orchestrates the full build-test-deploy loop autonomously.
Impact: Replaced manual coding workflows with a single command that produces reviewed, tested, deployed pull requests. Stop chatting, start shipping.
How This Was Built
Three independent tracks sharing a central data hub
Serving
Browser
End user
CloudFront + S3
Edge CDN
Next.js 16 SSG
App Router + Tailwind v4
Shared Data Hub
DynamoDB + S3
Posts · Subscribers · AI Evolution · Static assets
Daily Pipeline
EventBridge + Lambda
7 AM EST daily
GitHub Actions Curator
RSS scoring + AI Evolution + deploy
Resend
Newsletter email
Blog API
Blog Editor
GitHub OAuth
Blog API Lambda
CRUD + likes
Bedrock AI
Claude Haiku categorize
AI Evolution Farm
Weekly Curator
Categorize by topic
Evolution Updater
Top 5 per topic
AI Evolution Pages
13 interactive deep-dives
Static Export
Zero runtime cost. Pure HTML/CSS/JS served from edge CDN. No servers to maintain.
Claude Code
AI-assisted development using agent teams. Parallel agents built this site's Intelligence Hub in one session.
Daily Neural Digest
EventBridge triggers at 7 AM EST daily. Curates 20+ RSS feeds, sends email via Resend API, deploys site, and feeds AI Evolution weekly updates.
AI Evolution Farm
Interactive deep-dives into 10 AI topics with auto-curated weekly updates. Curator pipeline categorizes RSS entries by topic and surfaces top 5 advancements.
Blog API
Full CRUD blog with GitHub OAuth, likes, AI categorization via Bedrock, auto-save drafts. All serverless on Lambda + DynamoDB.