Anudeep Reddy Mutyala
Open to work8 years experienceAI Engineer
AI Engineer – Agentic Systems & LLM Applications
Standing Rank
Rank Not Available
Developer Badges
No badges earned yet
Skills & Technologies
31 skillsWork Experience
AI Engineer – Agentic Systems & LLM Applications
The Machinist (independent)
- Designed and deployed multiple agentic AI systems, evolving from raw execution loops to structured LangGraph pipelines
- Published detailed engineering case studies documenting architecture decisions, trade-offs, and system behavior
- Developed additional ML systems across manufacturing, automotive safety, and predictive maintenance
Founder & Head Trainer
A Bier Guy Consulting
- Implemented structured training programs with assessments for 200+ professionals, improving learning outcomes through iterative feedback
- Trained 50+ candidates to pass Cicerone certification exams on first attempt
- Earned Advanced Cicerone certification (1 of ~230 globally) through a 28-month structured preparation
Brewer – Production Operations
Windmills Craftworks
- Monitored fermentation variables (pH, gravity, temperature, dissolved oxygen) across 60+ batches; built Tableau dashboards supporting ~12% cost per hectoliter reduction
Technical Sales & Marketing Manager
Brewnation | Byg Brewski
- Built dashboards tracking pipeline, revenue, and product performance
- Improved conversion rate from ~35% to ~42% through segmentation and funnel analysis
- Managed workflows across 80+ SKUs and competitive intelligence tracking
Design Engineer – Underbody
Honda R&D Americas
- Supported $6–8M annual component spend decisions through structured data analysis and trade-off frameworks across 50+ underbody components
- Built KPI dashboards tracking design complexity, cost variance, and development velocity; identified 15–20% cost reduction opportunities through standardization
Projects
Content Agent – AI Research & Publishing System
View Project- Built a LangGraph-based content generation pipeline that retrieves sources, drafts articles, verifies factual claims, performs self-reflection, and publishes approved content to a live website through a two-stage human-in-the-loop workflow
- Designed a hybrid retrieval system combining BM25, dense vector search, and Reciprocal Rank Fusion across web search (Tavily) and a local knowledge base, with claim-level source grounding and verification — 100% recall@3 and 96.7% recall@1 on a 30-query golden eval set
- Implemented durable workflow execution using FastAPI and SqliteSaver, enabling interruption-safe human review, automated publishing, CI validation, rollback protection, and containerized deployment on AWS
Self-Correcting Code Agent
View Project- Built a LangGraph-based execution loop (execute → diagnose → patch → re-execute) with bounded reflexion loop (self-correction, max 5 iterations) and cost gating per iteration
- Achieved 95% fix rate on a 20-script benchmark — mean 2.3 iterations per fix, ~$0.0006 cost per run; only failure was a hard environment constraint (missing dependency in sandbox), not an agent logic error
- Stress-tested system with failure injection (timeouts, cost limits, invalid API keys) ensuring graceful recovery
Knowledge Agent – Hybrid RAG System
View Project- Designed hybrid retrieval pipeline: BM25 + dense search fused via Reciprocal Rank Fusion → cross-encoder reranking (ms-marco-MiniLM-L-6-v2); two-stage architecture (20 candidates → top 5)
- Achieved 92% accuracy and 100% tool routing accuracy on adversarial test cases
- Built claim verification tool with source grounding (chunk index + filename citation) and persistent JSON memory layer (last 10 queries + verified fact ledger)
- Evaluation harness and agent share the same imported system prompt — a deliberate constraint that eliminates silent divergence between test and runtime behavior
Silent Recall Detection System
- Automated ETL pipeline ingesting 10,000+ complaint and recall records from public APIs; risk metrics identified 28 vehicle models with complaint-to-recall ratios exceeding 50:1; ~90% reduction in manual investigation effort
Production Line Defect Analytics
- 1.18M+ records across 4,268 variables; feature space reduced ~80%; single station identified with 7.8× higher failure probability on 2.1% of throughput; storage optimized 60GB → 1.77GB Parquet
Bearing Failure Prediction & RUL Analytics
- 984 vibration recordings at 20,480 Hz; regression model R² = 0.985; late-stage prediction error reduced from ~30 hours to 2.88 hours via weighted loss optimization
Leaderboard Standings
Leaderboard Position Pending
Global test scores, peer standing percentiles, and algorithm leaderboard ranks are updated dynamically.
Assessment Highlights
Assessments Not Completed
Coding evaluations, system assessment results, and conceptual score badges will appear here after taking a test.
AI Collaboration Score
AI Collaboration Score Pending
Developer coding behavior, assistant cooperation, and AI pair-programming indicators are evaluated during live coding sessions.
Role Compatibility Profile
Role Compatibility Analysis Pending
Custom matching reports, candidate role compatibility percentiles, and core engineer strength profiles are processed once conceptual code screenings are complete.
Achievements
Advanced Cicerone Certification
Earned Advanced Cicerone certification (1 of ~230 globally) through a 28-month structured preparation.
About Details
Professional Bio
AI Engineer specializing in agentic systems and LLM-based applications, including multi-step pipelines, tool calling, RAG, and LangGraph state machines. Experienced in building end-to-end agent systems with benchmarking, regression testing, cost/latency tracking, and observability.
M.S. Mechanical Engineering
Wright State University, Ohio ( - Present)