Skip to content
View nagarjunak-pixel's full-sized avatar

Block or report nagarjunak-pixel

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
nagarjunak-pixel/README.md

Nagarjuna Reddy Kamireddy 👋

(Published on LinkedIn & GitHub as Nagarjuna Kamireddy · Nagarjuna K · Nagarjun Reddy Kamireddi)

Lead AI Architect  |  Production Agentic AI, Hybrid RAG & LLM Evals (750K+ Workflows/Yr)  |  Ex-John Deere, GSK, Deloitte  |  AWS SA Pro & NVIDIA Agentic AI Certified


Flagship Projects

  1. helios-consensus — Library-style Raft consensus in Python (~7.2K LOC Python + 705 LOC C++20 + ~1.8K LOC tests): PreVote, CheckQuorum, chunked snapshots, Ongaro §6 Joint Consensus (C_{old,new}), ReadIndex, crash-safe segmented binary WAL, asyncio TCP server/client & Multi-Raft range splits. Verified across 20/20 Jepsen fault-injection seeds (8,639 ops, Wing–Gong–Lowe linearizability) and 7/7 protocol/WAL mutant kills.
  2. tokendiet — Context compressor for AI coding agents: order-preserving knapsack pruning of tool logs (C++20 kernel + Python fallback), Python AST skeletons, cross-turn read dedup; CLI + MCP server. On 6 self-built tasks with a local qwen3.5:4b: 6,487 → 1,930 prompt tokens (70.25% fewer), required answer content kept in 6/6.
  3. slopguard — Rule-based checker for common AI-generated-code mistakes (stub bodies, unresolved imports, swallowed exceptions, tautological tests, dead UI handlers): Python AST + regex TS/JSX rules, C++20 prescreen, Claude Code PostToolUse hook & MCP server. ~0.03–0.28 ms per small file in the demo; 6-task local repair audit 0/6 → 6/6 passing its own rules.

Executive Impact (22+ Years Global Enterprise Data, SAP HANA & Production AI)

  • John Deere (AI Architecture & Technical Lead, 2023–2026): Led a 12+ member AI/ML team across 23+ production initiatives — 750K+ automated requests & 42K+ hours saved (~25% faster cycle time) in FY26. Built StoneAge (LangGraph + OpenSearch hybrid RAG + Neo4j copilot routing across GPT-5/Claude/GPT-4o/Gemini), rescued a failing ~30%-accurate GenAI product to ~70% accuracy via GPT-4o Vision + rule gates, built AIFlow (255 CI quality checks), and served on the AI Governance Council (NIST AI RMF / EU AI Act).
  • Persistent Systems, Vertiv & Soham Global (2016–2023): Production ML, MLOps, AWS Athena lakehouse pushdown, computer vision, NLP, and leakage-safe forecasting.
  • Global SAP HANA / BW / BPC Consulting (2004–2016 — UK, Malaysia, Singapore, US, India): GlaxoSmithKline (GSK UK) in-database HANA PAL/APL/R ML; PETRONAS Energy Trading & Deloitte (48 hours → 12 minutes HANA speedup); DHL, Symantec & Lenovo; 4× SAP Certified + Founder of Surya HANA Academy (~1,000 trained) & Suryagentic Institute (280+ live AI sessions in 2026).

LinkedIn Pulse Monographs & 15 Architecture Deep-Dives & 2 Scholarly Whitepapers


Systems & ML Reference Implementations (Python with small C++20 helpers)

  1. weavegraph — Code-graph scheduler for parallel agent worktrees: C++20 CSR graph, Tarjan SCC, min-cut task partitioning & 3-way AST merge check + MCP.
  2. vecta — Vector search engine: HNSW (0.999 recall, tombstones), IVF-PQ, RaBitQ-style 1-bit codes & hybrid BM25+RRF, with C++20 kernels.
  3. slopguard — Heuristic checker for AI-generated code: Python AST + regex TS/JSX rules, C++20 prescreen, Claude Code PostToolUse hook & MCP server.
  4. tokendiet — Context compressor for coding agents: order-preserving knapsack log pruning (70.25% fewer prompt tokens on 6 self-built tasks, 6/6 answers kept), AST skeletons & cross-turn dedup.
  5. syntara — Grammar-constrained structured decoding: JSON Schema/regex → minimal byte DFA (141.7x C++20 prefix-sharing index, 20/20 schema-valid) & byte-level Earley CFG.
  6. tracegit — Git-like local history for AI agent traces: SQLite/SHA-256 Merkle store, C++20 edit-distance diff/bisect, loop detection, CI gate & MCP server (prototype).
  7. aether-compiler — Educational tensor-graph compiler: SSA IR with CSE/fusion/DCE passes, C++20 static arena planner (432 → 120 MiB, 3.60x on a 2-layer example) & analytic H100 tile cost model.
  8. nexus-verity — Educational causal-inference toolkit: PC discovery (SHD=0 on a synthetic DAG), cross-fitted double ML (919x smaller bias than the naive estimate on synthetic data), split conformal & C++ doubly-robust OPE.
  9. helios-consensus — Library-style Raft (PreVote, Joint Consensus, ReadIndex), segmented binary CRC32C WAL, asyncio TCP + Multi-Raft sharding, Jepsen fault-injection harness & C++20 Wing–Gong–Lowe linearizability solver (20/20 seeds safe, 7/7 mutants killed).
  10. privora-graph — Privacy & robustness reference implementation: DP-SGD with RDP accounting, bi-temporal PageRank retrieval, Multi-Krum/geometric-median aggregation.
  11. quant-forge — LLM weight-quantization reference implementation: GPTQ, AWQ, Walsh-Hadamard rotation, simulated FP8 & a small C++20 packed-INT4 GEMM.
  12. veritas-rl — RL post-training sandbox: GRPO/DAPO on toy math tasks (6.25% → 50.00% Pass@1), DoRA vs LoRA on a 128×128 toy (563x lower MSE), single-process ring-attention/TP simulation & step-level PRM reranking.
  13. axiom-zero — Educational small GQA transformer: speculative decoding with a C++ helper (4.00x fewer target passes), Top-K SAE steering (R²=0.9855) & CDCL SAT solver with RUP checks.
  14. aegis-runtime — Policy and audit layer for AI-agent tool calls: custom policy DSL, taint tracking, PII vault, saga compensation & hash-chained Merkle/Ed25519 audit log.
  15. prism-gate — OpenAI-compatible LLM gateway prototype (~4,660 req/s gateway overhead against simulated upstreams): TCP-Vegas adaptive concurrency, heuristic cascade routing & PSI/JSD drift checks.
  16. chronos-mesh — Educational supply-chain rerouting model: pure-Python simplex + branch-and-bound MILP, BOM safety stock & Monte Carlo CVaR95.

Complete Public Repository Index (Interactive Atlas →)

Repository Stack Architectural Scope
nagarjunak-pixel Multi-Lang Nagarjuna Reddy Kamireddy (Nagarjuna Kamireddy / Nagarjuna K) — Lead AI Architect, Agentic AI, GraphRAG, LLM Evaluation & SAP HANA
nagarjunak-pixel.github.io HTML Nagarjuna Reddy Kamireddy (Nagarjuna Kamireddy / Nagarjuna K) — Official Lead AI Architect, Agentic AI & SAP Portfolio, Case Studies & llms.txt Hub
weavegraph Python Code-graph worktree scheduler, min-cut multi-agent task partitioner & 3-way AST merge check (C++20 + Python + MCP + React)
calibra (v1.0.0) C++20 · Python · mypy --strict (0 errors) · 99.6% coverage (>100 tests) Rigorous eval: bootstrap/McNemar tests, Bradley–Terry leaderboards, 2PL IRT anchor benchmarking, ECE temperature scaling, and MinHash-LSH contamination detection. Bit-exact C++20 kernels.
kvserve (v1.0.0) C++20 · Python · mypy --strict (0 errors) · 99.6% coverage (>100 tests) Paged-KV LLM serving engine: PagedAttention block manager with refcounted automatic prefix caching, Orca continuous batching, Sarathi chunked prefill, recompute preemption, and bit-exact C++20 kernels.
vecta Python Vector search engine: HNSW (heuristic neighbour selection, filtered & tombstone search) at 0.999 recall@10 touching 1.4% of vectors, IVF-PQ ADC + rerank (24x compression), RaBitQ-style 1-bit codes, BM25+RRF hybrid & bit-exact native C++20 kernels
slopguard Python Heuristic checker for AI-generated code — Python AST + regex TS/JSX rules for stubs, unresolved imports, empty tests; C++20 prescreen, Claude Code hook, MCP
tokendiet Python Context compressor for AI coding agents — knapsack log pruning (C++20 kernel, Python fallback), AST skeletons, cross-turn read dedup. CLI + MCP server
nrk-enterprise-ai-masterbook Multi-Lang Nagarjuna Reddy Kamireddy™ — The 7,200-Page Enterprise AI, Claude Code & SAP RAG Masterbook Series (50 Daily Chapters)
syntara Python Grammar-constrained structured decoding: JSON Schema/regex -> minimal byte DFA (Thompson/Rabin-Scott/Moore), C++20 prefix-sharing token index (112-142x, bit-exact vs Python), byte-level Earley CFG constraints & closed-set choice scoring
tracegit Python Git-like local history for AI agent traces — SQLite/SHA-256 Merkle store, C++20 edit-distance diff/bisect, loop detection, CI gate, MCP server. Prototype
nexus-verity Python Causal inference in Python — PC discovery, d-separation, cross-fitted double ML, split conformal, C++ doubly-robust OPE. Educational reference implementation
aether-compiler Python Small tensor-graph compiler: SSA IR with CSE/fusion/DCE passes, C++20 static arena planner, analytic H100 tile cost model. Educational reference implementation
privora-graph Python Privacy & robustness toolkit — DP-SGD with RDP accounting, bi-temporal PageRank retrieval, Multi-Krum/geometric-median aggregation. Reference implementation
helios-consensus Python Library-style Raft consensus with segmented binary WAL, asyncio TCP, Multi-Raft range splits, Jepsen fault injection, Wing–Gong–Lowe linearizability & C++20 kernels
venkatesh-llm HTML Micro-LLM From Scratch (1.03M Parameters) — End-to-end BPE tokenizer, causal Transformer pretraining, checkpointing & Ollama Q&A engineering lab.
surya-agentic-academy-digest HTML Automated Agentic AI Research & Engineering Digest — Daily LLM paper & technical talk synthesis into 5 runnable project blueprints per edition.
langgraph-localai-tv Python LangGraph 3-stage news pipeline: headlines → video → bulletin. Human approval gates, checkpoint persistence, error recovery.
self-improving-agents Python 🧠 Self-Improving Multi-Agent System — A teaching framework for advanced agentic AI: ReAct loop, Reflexion, multi-agent orchestration, adversarial reasoning, and agent-as-a-judge. With web UI + Netlify deployment.
operdai-dashboard JavaScript Enterprise Multi-Tenant White-Label AI Lead-Agent Analytics Console — React + TypeScript telemetry, conversion funnels & partner branding portal.
loop-engineering-trading JavaScript Browser 'Minara' quant desk: research, strategy, execution, and review loops with backtesting and paper trading.
llm-knowledge-base Python LLM knowledge base: ingest sources, compile a markdown wiki, search, and ask questions (FastAPI + SQLite).
firecrawl-nexus-rss Python Firecrawl-powered news aggregator: scrapes AP/TG constituency news into clean markdown and structured JSON.
deerflow-research-agent Python DeerFlow-inspired research agent with sub-agents (Planner, Searcher, Analyzer), persistent memory, and session resume.
crewai-telugu-newsroom Python CrewAI multi-agent Telugu newsroom: Researcher → Script Writer → Fact-Checker. Sequential pipeline for automated news production.
codexforge-tdd-mvp-v2 Python Self-correcting TDD sandbox: RED→GREEN loop with Docker isolation, VS Code extension, and LLM-powered code repair
apsrtc-statistics-dashboard JavaScript React dashboard for APSRTC operations: fleet, tickets, financials, route heatmap, EV tracker, and a what-if simulator.
agentic-ai-dashboard JavaScript Interactive Visual Guide to Agentic AI Architecture — Tokenizer Inspector, RAG Playground, ReAct Loop, Reflexion & LLM Evaluation Suite.
agentic-academy Python Python teaching framework for agentic engineering: agents, a harness, loops, memory, and a local-model demo.
agentforge JavaScript AgentForge — Interactive React Workbench for Designing, Simulating & Debugging Multi-Agent Graphs, Eval Sandboxes & Swarm Telemetry.
agent-framework-lab Python Hands-on Microsoft Agent Framework 1.0 lab — multi-agent workflows on local Ollama (no API keys)
pytorch-interactive-demo JavaScript Interactive PyTorch Deep-Learning Visualizer — Vector Embeddings, Cosine RAG Retrieval & Autograd Training Loop Dynamics.
attention-head-walkthrough HTML Interactive Step-by-Step Transformer Self-Attention Head Simulation (Q/K/V Projections, Causal Masking & Softmax Attention Weights).
ai-calc JavaScript Qwen AI Calculator — plain-English math in the browser, solved locally with Ollama.
company-os Python CompanyOS — config-driven team of AI agents for simulated company workflows (company.json, 7 role-agents, SIM + LIVE modes, kanban dashboard, KPI engine).
rag-chat-docs Python RAG Chat with Documents — 7-Module From-Scratch Retrieval-Augmented Generation System (Chunking, Hybrid Search, CLI & FastAPI Web UI).
spring-boot-mcp TypeScript MCP server that lets an LLM inspect a Java Spring Boot app via Actuator, REST APIs, and logs.
localaitv-mcp-server Python Python MCP server for LocalAI TV admin access to Postgres, AWS (ECS, S3, SQS, CloudWatch), and the public API.
antigravity-sdk Python Example agents for Google's Antigravity SDK: chat, custom tools, MCP, streaming, multimodal, and hooks.
ai-workforce Python Local multi-agent workforce with a FastAPI dashboard: CEO → research → architect → coding → QA → docs → deploy → marketing, on Ollama.
smart-marketing-assistant Python CrewAI + LangChain marketing crew on local Ollama that drafts a content strategy, plus a daemon that can post AI news to X.
agentic-systems-v3.1 Python Enterprise Agentic Systems v3.1 — consolidated monorepo (Phases 0-7)
surya-shiksha TypeScript Surya Shiksha — Multi-Agent Adaptive Learning Operating System (12 Specialized AI Agents, Next.js, TypeScript, PostgreSQL & Prisma).
quant-forge Python LLM weight-quantization reference implementation — GPTQ, AWQ, Walsh-Hadamard rotation, simulated FP8, and a small C++20 packed-INT4 GEMM. Educational.
veritas-rl Python RL post-training sandbox — GRPO/DAPO on toy math tasks, DoRA, single-process ring-attention/TP simulation, step-level PRM reranking. Reference implementation
axiom-zero Python Small GQA transformer with speculative decoding (C++ helper), Top-K SAE steering, and a CDCL SAT solver with RUP checks — educational reference implementation
chronos-mesh Python Supply-chain rerouting toy model — pure-Python simplex + branch-and-bound MILP, BOM safety stock, Monte Carlo CVaR95. Educational reference implementation
prism-gate Python OpenAI-compatible LLM gateway prototype — TCP-Vegas adaptive concurrency, heuristic cascade routing, PSI/JSD drift checks. Reference implementation
aegis-runtime Python Policy and audit layer for AI-agent tool calls — custom policy DSL, saga compensation, hash-chained Merkle/Ed25519 audit log. Reference implementation
02-autogpt-autonomous-agent Python Forward-Deployed Engineering (FDE) Autonomous Agent Lab — Zero-dependency ReAct think-act-observe execution loop & enterprise FDE curriculum.
nagarjuna-kamireddy-ai-architect-dataset JSONL Open ChatML/SFT Fine-Tuning & RAG Grounding Dataset for Nagarjuna Reddy Kamireddy

Direct Contact

Pinned Loading

  1. company-os company-os Public

    CompanyOS — run ANY company with a team of AI agents. Config-driven (company.json), 7 role-agents, SIM + LIVE modes, kanban dashboard, KPI engine.

    Python

  2. llm-knowledge-base llm-knowledge-base Public

    LLM knowledge base: ingest sources, compile a markdown wiki, search, and ask questions (FastAPI + SQLite).

    Python

  3. surya-shiksha surya-shiksha Public

    Surya Shiksha — Multi-Agent Adaptive Learning Operating System (12 Specialized AI Agents, Next.js, TypeScript, PostgreSQL & Prisma).

    TypeScript

  4. rag-chat-docs rag-chat-docs Public

    RAG Chat with Documents — 7-Module From-Scratch Retrieval-Augmented Generation System (Chunking, Hybrid Search, CLI & FastAPI Web UI).

    Python

  5. langgraph-localai-tv langgraph-localai-tv Public

    LangGraph 3-stage news pipeline: headlines → video → bulletin. Human approval gates, checkpoint persistence, error recovery.

    Python

  6. self-improving-agents self-improving-agents Public

    🧠 Self-Improving Multi-Agent System — A teaching framework for advanced agentic AI: ReAct loop, Reflexion, multi-agent orchestration, adversarial reasoning, and agent-as-a-judge. With web UI + Netl…

    Python