profile = {
"name" : "Srikanth Reddy Nandireddy",
"role" : "Applied AI / LLM Engineer",
"education" : "M.S. Data Science & AI — University of Central Missouri",
"gpa" : 4.0,
"location" : "United States",
"open_to_work" : True,
"available" : "Immediately",
"focus" : ["Multi-Agent Orchestration", "LLM/RAG Systems", "Agent Safety", "MLOps"],
"frameworks" : ["Microsoft Agent Framework", "MCP", "ReAct", "FastAPI"],
"recent_win" : "HackerRank Orchestrate — 16 / 1,773 (Top 1%)",
"portfolio" : "https://srikanthreddynandireddy.me",
"gfg_streak" : "255 consecutive days",
}I build agentic AI systems that ship — multi-agent pipelines with Critic/Verifier
loops, MCP-grounded retrieval, layout-aware RAG, and the MLOps automation that takes them
from notebook to a live /predict endpoint you can hit with curl.
|
5-agent Microsoft Agent Framework system — Planner-Executor + Critic/Verifier loop-back that flips Not Ready → Ready, MCP-grounded citations, all three Microsoft IQ layers. 1.0 readiness/risk accuracy on the ground-truth set.
|
Two-stage VLM agent for insurance-claim adjudication, injection-safe by design (7/8 attacks blocked). Ranked 16 / 1,773 — Top 1% at HackerRank Orchestrate. 95% valid-image accuracy at ~$0.017/claim.
|
|
LLM medication-logging agent with a deterministic safety spine — the model is load-bearing at exactly two edges, everything between is deterministic Python. Writes are gated by a single-use signed JWT (unique JTI, 300s TTL, subject binding), so a compromised model provably cannot log an off-schedule dose.
|
Cognitive ML research agent: multi-hop ReAct planning, layout-aware RAG, adversarial double-blind peer review, recursive citation tracing. MRR@5 0.990 — 7.3× answer relevance over BM25.
|
|
Production MLOps system — automated CI/CD retraining, MLflow Model Registry, Dockerized FastAPI on AWS EC2 behind Nginx, with
|
SARIMAX vs Prophet vs XGBoost on 3,300+ records. Holdout MAE 138.14 ≈ CV MAE 137.86 — near-zero overfitting. Plotly Dash dashboard on Render.
|
|
BERT + Wav2Vec2 + ViT cross-modal Transformer fusion (4 layers, 8 heads). Weighted F1 60.66% on 7-class MELD, class-weighted loss for imbalance.
|
TF-IDF + Logistic Regression vs DistilBERT on 62K+ articles, with dynamic F1-threshold tuning to trade precision against recall.
|
Also on the bench: 📊 arXiv RAG Benchmark — empirical study of 4 retrievers × 3 chunk sizes over 150 papers
Languages
Agents & LLM Systems
ML & Deep Learning
Cloud & MLOps
| Award | Details |
|---|---|
| 🥇 Graduate Student Achievement Award 2025–2026 | University of Central Missouri Graduate Studies — Recognized for exceptional scholastic achievement, exemplary contributions to Data Science & AI, and outstanding character and leadership |
| 🏅 HackerRank Orchestrate — Top 1% | Ranked 16 / 1,773 · 24-hour AI agent hackathon (June 2026) |
| 🥇 Global Rank 1 | CodeChef May Long Two — Division 4 Rated |
| 🏆 National Finalist | IICC Coding Competition — Top 1% of 100,000+ participants |
| 🥉 Bronze Medal | Ranked Top 3 in B.Tech Academic Program |
| ⭐ 5★ HackerRank | Data Structures & Algorithms |
| 🔥 255-Day Streak | GeeksforGeeks Problem of the Day |
| Degree | Institution | GPA |
|---|---|---|
| M.S. Data Science & AI (2024–2026) | University of Central Missouri | 4.0 / 4.0 |
| B.Tech CSE (Hons.) (2020–2024) | CMR College of Engineering & Technology | 9.06 / 10.0 |
Certifications: Microsoft Applied Skills — Developing Agents in Microsoft Foundry · Anthropic — Model Context Protocol · Anthropic — AI Fluency · AWS Academy — Data Analytics & Cloud Foundations · Google Data Analytics
I'm actively seeking full-time Applied AI / LLM Engineer & ML Engineer roles in the United States — available immediately.
