Available for select rolesAI ENGINEER · INDIA
Gopalakrishna
Maddipalli
I build production-grade AI systems across RAG, multi-agent workflows, model evaluation, and cloud deployment.
Available for select rolesGopalakrishna
CURRENT FOCUS
Systems that reason
with context.
Evaluation-led RAGStateful agentsLow-latency serving
LIVE PIPELINEOnline
> retrieving context… ▍
Evidence-aware portfolio
10
documented systems · targets labelled
Review evidenceFEATURED SYSTEM
Autonomous Work Order Intelligence & Operations Platform
Microsoft FoundryPythonFastAPI
EXPERIENCE
7+
Years across construction, data & AI
SunRay Construction Solutions · Hyderabad, India
WRITING
Engineering notes
Typed decisions, production failures, and reliable AI systems.
CORE TOOLKIT
PythonFastAPILangGraphMCPNext.js
COLLABORATE
Let's build something reliable.
Open to select Generative AI engineering and architecture roles.
01 / Profile
I build AI systems that move from prototype → production.
Seven years across construction operations and data science inform a workflow-first approach to Generative AI, RAG, and autonomous agents.
GopalakrishnaIndia · Available
I focus on prompt engineering, advanced RAG topologies, autonomous multi-agent workflows, and the evaluation systems required to make them reliable — spanning Data Science, Generative AI, LLM applications, Machine Learning, and AWS.
Viewing as
Available for Generative AI Engineering and AI Architecture roles, based in India — 7+ years domain experience, 10 blueprint systems, 28 credentials retained.
CurrentAI/ML Engineer · SunRayDomainConstruction legal operationsEvidenceRésumé · LinkedIn · credentials
7+
Years domain experience
10
Blueprint systems
28
Credentials retained
Current focus
Evaluation-led RAG
Stateful agent orchestration
Low-latency model serving
System principle / 01
I build systems that think with context.
Portfolio film / Ten connected systems
From workflow friction
to governed intelligence.
Payment protection
Construction Payment Risk Prediction
Risk signals → human reviewSynthetic workflow intelligence · frozen threshold 0.20Space: play/pause · R: restart · F: fullscreen
Synthetic where stated · Source-aware · Human-reviewed · No automated legal decisions
02 / Selected work
Selected AI systems.
Ten AI systems with explicit goals, implementation stacks, and system flows. Figures marked as targets are project targets — not unverified production claims.
TTFT target · <150msOrchestration · LangGraphVector stores · Qdrant / FAISS
Flagship system · Case 02
Evidence publishedAutonomous Work Order Intelligence & Operations Platform
Problem
An evidence-aware, multi-agent architecture for construction payment-protection operations — processing work orders through specialist agents for intake, research, evidence, and discrepancy detection, with controlled correction and human-in-the-loop escalation for complex cases.
Engineering
Microsoft Foundry · Python · FastAPI · MCP · AWS · Azure Container Apps
Impact
~99% target evaluation baseline
Agent coordination
ComplianceRiskCommunication
A planner routes each request to specialist workstreams, calls tools where needed, and holds the result for human approval before a final response is returned.
Flagship system · Case 07
In developmentAI Legal Assistant
Problem
Answer construction-law questions with grounded, source-aware retrieval.
Engineering
LangChain · AWS Bedrock · FAISS · FastAPI
Impact
95% retrieval accuracy target
Retrieval loop
Documents are embedded and indexed ahead of time; at query time the retriever pulls and reranks the closest matches before the LLM generates a grounded, source-aware answer.
01 / 10
Machine Learning
Evidence publishedConstruction Payment Risk Prediction
Human-reviewed operational prioritization
Prioritize construction payment-protection workflows before delay or escalation becomes harder to manage.
PythonFastAPIAWSDataRobotSHAP
System flow
S3 → Glue → Athena → SageMaker → ECS → CloudWatch
03 / 10
Agentic AI
Evidence publishedNTO Operations Copilot
Deployed evidence-grounded research coach
Guide new Notice to Owner researchers through the complete operational process, retrieve work-order evidence, explain exceptions, and recommend the next approved action without replacing human verification.
Microsoft FoundryMCPFastAPINext.jsAWS LambdaAzure App Service
System flow
Work order → Intake evidence → Guided research → Notice selection → Human review
04 / 10
Document AI
In developmentConstruction Document Intelligence
Reduced manual document processing
Extract structured, reviewable information from complex legal PDFs.
TextractPaddleOCRspaCyPostgreSQLFastAPI
System flow
PDF → OCR → NER → JSON → Database
05 / 10
Machine Learning
In developmentLien Recommendation Engine
Faster compliance decisions
Recommend notices, deadlines, and the next appropriate legal action.
PythonMachine LearningRules EngineFastAPI
System flow
Project data → Risk model → Rules → Recommendation
06 / 10
Data Engineering
In developmentConstruction Data Lakehouse
Unified reporting foundation
Create a governed source of truth for operational and executive analytics.
PySparkAWS GlueAthenaRedshiftPower BI
System flow
CRM → S3 → Glue → Athena → Dashboards
08 / 10
Document AI
In developmentClaims & Contract Intelligence
Accelerated contract review
Analyze claims and contracts for obligations, risks, and missing evidence.
OCRTransformersRAGAWS
System flow
Contract → Extraction → Risk analysis → Evidence
09 / 10
Machine Learning
In developmentExplainable Payment Intelligence
Transparent risk decisions
Turn payment predictions into clear, defensible business explanations.
XGBoostSHAPLLMsPower BI
System flow
ERP → ML → Explanation → Dashboard
10 / 10
Agentic AI
In developmentConstruction Compliance Copilot
Auditable human-AI collaboration
Deliver end-to-end assistance across documents, deadlines, and communication.
LangGraphMCPCrewAIFastAPI
System flow
Intake → Plan → Tools → Review → Audit trail
Evidence index
Published proof, without repeated project browsing.
Open the three projects with published evidence here. The remaining 7 systems are labelled in development within Selected Work.
3published7in development
Machine Learning
Construction Payment Risk Prediction
Published evidenceCase study and available artifacts
InspectAgentic AI
Autonomous Work Order Intelligence & Operations Platform
Published evidenceCase study and available artifacts
InspectAgentic AI
NTO Operations Copilot
Published evidenceCase study and available artifacts
Inspect03 / Interactive lab
See how prompt structure changes an answer.
A live playground calling a real model through this site's own API — adjust temperature and top-p and inspect the actual response. Falls back to a static example if the live model is unavailable.
Outputs · live modelTiming · measured server-sidePurpose · interaction study
Loading technical module…
Technical lab · Infrastructure observability
Loading technical module…
Technical lab · Production RAG pipeline
Loading technical module…
04 / Experience
Construction knowledge, developed into applied AI.
A verified progression from CAD and construction research through data science to AI/ML engineering at SunRay Construction Solutions.
SunRay Construction Solutions
One company. Three chapters of increasing technical responsibility.
Hyderabad, Telangana, India · Full-time
AI/ML Engineer
Build and implement AI-powered applications for construction legal operations, research, and compliance workflows.
Data Scientist
Applied data science and machine learning to construction operations, research, and payment-protection workflows following completion of postgraduate study in AI and machine learning.
Research Analyst
Supported construction legal research and operational work-order processing with emphasis on accurate source verification and documented decision-making.
MAY 2019 — SEP 2019
Venusgeo Solutions
Foundation role
Junior CAD Engineer
India · 5 months
Contributed to a UK county-council asset-management project by producing accurate building and school-facility information.
05 / Capabilities
The execution stack.
01
Models & Fine-Tuning
Llama 3QwenDeepSeekLoRA / QLoRAAxolotlUnslothHugging FaceGPT-5-miniDataRobot AutoMLSHAPscikit-learn
02
Inference & Optimization
vLLMTensorRT-LLMOllamaAWQGGUFGPTQFlashAttention
03
Orchestration & Agents
LangChainLangGraphAutoGenCrewAILlamaIndexDSPyMicrosoft FoundryMCP
04
Vector Databases & RAG
QdrantPineconeMilvusChromaHybrid SearchRRF
05
Engineering & Infrastructure
PyTorchPythonFastAPIDockerRayTritonCUDAAWS / GCPAWS S3AWS GlueAmazon AthenaAWS IAMAmazon SageMakerAmazon ECSAmazon CloudWatchAWS LambdaAPI Gateway
06
Frontend & Deployment
Next.jsReactTypeScriptTailwind CSSAzure Container AppsGitHub ActionsOIDCpytest
06 / Verified badges
Credentials you can check, not take on trust.
Digital badges issued through Credly. Each one is tied to the issuer's own record, so the claim can be verified independently of this site.
Python Project for AI and Application DevelopmentIBM · CourseraIssued Jul 22, 2026
07 / Credentials
Formal training behind the practice.
Primary · UT Austin / Great LearningCredential 02Credential 03Credential 04Credential 05Credential 06
Post Graduate Program in Artificial Intelligence and Machine Learning
McCombs School of Business, The University of Texas at Austin · Great Learning · Jan 2021
Agent Architect
Founderz AI & Business School · Certificate of Completion · Sep 2026
Agent Explorer
Founderz AI & Business School · Certificate of Completion · 2026
Generative AI for Data Scientists
IBM · Coursera Specialization · Sep 2026
Generative AI for Data Analysts
IBM · Coursera Specialization · Sep 2026
Generative AI: Elevate your Software Development Career
IBM · Coursera · Sep 2026
What I bring
Technical work grounded in domain experience.
01
I understand the workflow
Seven years in construction operations taught me where delays, ambiguity, and risk actually enter a process.
02
I work comfortably with data
My foundation includes predictive modelling, explainability, document processing, and practical analytics.
03
I design for responsible use
I treat evaluation, human review, cost, latency, and failure handling as part of the product—not afterthoughts.
08 / Writing
Notes on system design.
Agentic architecture, governed AI, and production ML, written from the systems I build.
3 Oct 2026 · 7 min read2 Oct 2026 · 5 min read2 Oct 2026 · 6 min read30 Sept 2026 · 6 min read
Your AI Eval Suite Can Become Technical Debt
Turning every production failure into a permanent eval eventually makes the suite the bottleneck. Triage first: deduplicate the known, protect the novel, tier what runs where, and evaluate the triage too.
A System Prompt Is Not a Security Boundary
Agents can now query databases, call APIs and run code. What limits them should live in identity, policy, sandboxing and tool authorization, not in an instruction to the model.
Your Agent's Most Valuable Dataset Is Its Production Failures
CoreWeave Forge frames AI work as a run-observe-curate-improve-evaluate loop. The engineering lesson: map each failure type to an eval, test every layer of an agent, and gate every fix.
Typed Decisions, Not Paragraphs: What Jev Gets Right About Agentic Systems
Why decision steps in agent systems need typed outputs, calibrated confidence, and a human exit, and how TypeSafe AI's Jev model approaches that from the model side.
09 / Contact
Let's build intelligent systems.
Best-fit collaborations
Evidence-grounded RAG, agentic workflows, construction-domain AI, and production evaluation.
Open to thoughtful conversations about Generative AI engineering, AI architecture, and applied research.
Download résumé3D animation intensity set to Balanced