Open to Enterprise AI Architecture and Senior GenAI Roles

Rishikesh
Pote

|

7.5+ years of experience building enterprise data platforms and production AI systems — from scalable ETL foundations to end-to-end agentic automation. I design multi-agent orchestration, retrieval pipelines, evaluation harnesses, and governance controls for regulated analytics workflows.

View Case Studies LinkedIn
7.5+ Years Exp.
3 Case Studies
75% Efficiency Gains
8+ Certifications
Multi-Agent Orchestration LLM Evaluation Systems Guardrails and Auditability
ai_architect.py 
# Rishikesh Pote - AI Engineering Architect

class RishiPote:
    focus   = "Agentic AI · RAG · EvalOps"
    journey = "6yr Data Engineering → 1.5yr AI Architecture"

    ai_stack = [
        "LangGraph", "LangChain",
        "LangSmith", "Vector RAG",
        "LLM-as-Judge", "Model Routing",
    ]

    platform = [
        "Databricks", "AWS",
        "PySpark", "Snowflake",
    ]

    def build(self):
        return "Spec → Code → QC → Audit-ready Output"

    def philosophy(self):
        return "Deterministic core · Measured AI autonomy"
LangGraph
AWS Certified
RAG Systems
01 / About

Who I Am

RP

I am a Senior AI/Data Architect and hands-on GenAI Engineer focused on agentic automation for regulated analytics workflows. Across 7.5+ years of experience — including 6 years in enterprise data engineering and 1.5 years in AI architecture — I have developed both the technical depth and architectural discipline needed to ship production-grade AI systems.

My AI engineering work centers on building end-to-end LLM systems: retrieval pipelines, multi-agent orchestration (LangGraph/LangChain), LLM-as-Judge evaluation frameworks, and governance patterns on cloud-scale data platforms. I convert ambiguous domain processes (protocol/spec → code → QC) into repeatable AI pipelines with measurable quality gates — with a strong emphasis on correctness, traceability, and enterprise integration.

Before AI Engineering, I spent 6 years as a Senior Data Engineer & Cloud Solution Architect, architecting healthcare data platforms on AWS & Azure with PySpark, Databricks, Snowflake, and Kafka. I hold a B.E. in Computer Science from LIT, GTU (CGPA 8.96), and multiple AWS and Databricks certifications.

Pune, Maharashtra, India
ZS (Current)
B.E. Computer Science, LIT GTU
AWS SA + AI Practitioner Certified
Databricks Certified (3×)
GenAI · LangGraph · RAG · EvalOps
Engineering Philosophy
Spec-first Workflows Structured outputs as contracts for downstream code & QC
Closed-loop Improvement Critic + LLM-as-Judge signals drive iterative refinement
Deterministic Kernel Rules for safety/consistency; LLM for flexibility
Confidence Thresholds Auto-accept vs mandatory review vs high-priority escalation
Observability Mindset Run metadata, audit trails, error recovery, rollback plans
02 / Experience

Work History

AI Engineering Architect & Senior GenAI Engineer

ZS
2024 – Present Pune, India
AI Engineering Track
  • Architect and deliver end-to-end agentic pipelines for regulated clinical analytics — SDTM → ADaM and Protocol PDF → Spec → R Code → QC workflows.
  • Design multi-agent orchestration systems (LangGraph/LangChain) using Supervisor-Worker, Critic-Reviewer, and Router patterns with guarded execution and idempotency.
  • Build RAG knowledge layers with chunking strategy, metadata-aware retrieval, citation-first answer shaping, and confidence-based routing.
  • Implement LLM-as-Judge evaluation frameworks across structural QC, logic QC, and scorecards aligned to gold standards for 10+ studies.
  • Define architecture guardrails and governance artifacts (HLD/LLD, repository standards, observability baselines) for enterprise GenAI platforms.
LangGraphLangChainRAGFAISS LLM-as-JudgeDatabricksClaudeOpenAI

Senior Data Engineer & Cloud Solution Architect

ZS
2020 – 2024 Pune, India
  • Architected cloud-native lakehouse solutions for enterprise healthcare clients on AWS & Azure — compute, storage, secrets, and search layers at scale.
  • Designed scalable ETL/ELT pipelines using PySpark, Databricks, and Snowflake handling millions of records with Git-based SDLC, CI/CD, and environment parity.
  • Built API gateway patterns for key authorization/routing with full audit logging and operational metadata for end-to-end traceability.
  • Led Customer Engagement Framework implementations involving batch & streaming processing on Databricks and Kafka for healthcare data platforms.
  • Mentored junior engineers; drove internal standards for data architecture, release hygiene, and code quality across cross-functional teams.
PySparkAWSAzureDatabricks SnowflakeKafkaHealthcare

Data Engineer

Exponentia.ai
2019 – 2020 India
  • Developed configuration-based big data ETL frameworks on Cloudera, AWS & Azure, eliminating manual processing dependencies across pipelines.
  • Built an Automated Python Framework that significantly reduced pipeline run times and removed manual intervention points.
  • Delivered the Customer Engagement Framework project involving PySpark and streaming data across cloud environments.
  • Recognised with the Kudos Award (March 2020) for outstanding project performance and delivery.
PythonPySparkCloudera AWSAzureETL
03 / Skills

Tech Stack

LangGraph LangChain Vector RAG FAISS Chroma LLM-as-Judge Agentic AI Gradio Streamlit Claude API OpenAI API Tool Use LangGraph LangChain Vector RAG FAISS Chroma LLM-as-Judge Agentic AI Gradio Streamlit Claude API OpenAI API Tool Use
Python PySpark AWS Azure Databricks Snowflake Apache Kafka R Git / CI/CD Data Governance API Gateway Lakehouse Python PySpark AWS Azure Databricks Snowflake Apache Kafka R Git / CI/CD Data Governance API Gateway Lakehouse

Agentic AI & Orchestration

LangGraphLangChainSupervisor/Worker
Critic LoopsTool InvocationRouter Patterns

RAG & Knowledge Layer

FAISSChromaVector Search
Chunking/OverlapMetadata RetrievalEvidence Shaping

LLM Evaluation & QC

LLM-as-JudgeStructural QCLogic QC
Eval MatrixConfidence ScoringHITL Feedback

Spec & Code Automation

Protocol → SpecSpec → CodeR Programming
Debugger AgentsQC ReportsSDTM / ADaM

Data Engineering (Foundation)

PySparkDatabricksKafka
SnowflakeAWSAzure

Platform & DevOps

Lakehouse Arch.API GatewayCI/CD
Audit LoggingSecrets Mgmt.Git SDLC
04 / Case Studies

Enterprise AI Engineering

Real-world agentic systems built for regulated, enterprise-grade workflows. Client names intentionally withheld.

Large Pharma Enterprise

Clinical GenAI: SDTM → ADaM Automation

AI Engineering Architect / Senior AI Engineer
Production

End-to-end agentic pipeline automating SDTM → ADaM dataset generation and QC reporting at enterprise scale.

16 ADaM Domains ~10 Studies ~75% Efficiency Gain
Agent Pipeline
Spec Generation
Code Generation
RAG Grounding
LLM-as-Judge QC
Debugger / QC
Final Report
Results
16 ADaM Domains
~10 Studies
45–60 min Runtime
~75% Efficiency Gain vs Manual
Deliverables: HLD/LLD + Architecture Diagrams Repo Structure + Guardrails Stakeholder Walkthroughs
LakehouseVector RAGGit SDLC API GatewayClaudeOpenAI
Consulting Engineering Track

Clinical GenAI: Raw → SDTM Automation

SME / AI & Data Engineering Architect

Architecture and engineering for automating Raw → SDTM transformations in a Databricks lakehouse pipeline.

Pipeline Design LLM Code Review
Pipeline design + orchestration
Toolchain integration
LLM-powered code review utility
Repo conventions + maintainability guardrails
DatabricksLangChainLLM Code Review PySparkSDTM
Large Pharma Enterprise · Ongoing

RWD Agentic System: Protocol PDF → Spec → Code → QC

AI Engineering Architect / Lead

Automates the full analyst workflow from Protocol PDF (30–40 pages) through spec, R code, and QC validation.

15 RAG Pairs Eval Matrix
Authoring Agent (Protocol → Spec + Critic/Review loop)
QC Agent (Structural + Logic QC, analyst-reviewable reports)
RAG Knowledge Layer (~15 historical protocol/spec pairs)
Evaluation Matrix Engine (metric trend tracking)
LangGraphStreamlitVector RAG REval MatrixSQL Audit
05 / Projects & Demos

What I've Built

Personal AI / GenAI Demos
Personal

Dr. Assistant — RAG Copilot

Local-first RAG chatbot with Gradio dark-theme UI, sources table, and latency display.

FAISSGroq LLMGradio

PDFs → chunk/embed → FAISS → LLM answers with sources & latency display. Gradio dark-theme UI with sources table (file / page / chunk / preview) and loaded-PDFs panel.

FAISSsentence-transformersGroq LLMGradio
Personal

Local RAG App — Ollama + Chroma

Privacy-first local RAG pipeline — no data leaves the machine.

OllamaChromaStreamlit

Privacy-first local RAG pipeline: Ollama LLM + embeddings, persistent Chroma vector DB, and a clean Streamlit UI. No data leaves the machine — built for secure, offline document Q&A.

OllamaChromaStreamlitLocal LLM
Architecture Design

Manufacturing RAG — Knowledge + Actions + Audit

Sellable architecture for manufacturing: RAG + workflow agents with safety controls.

RAGAgentsRBAC

Sellable architecture for the manufacturing domain: RAG knowledge layer + workflow agents with safety controls, RBAC + action allowlists, doc revision prioritization, evidence thresholds, and full audit logs.

RAGAgentsRBACAudit Logs
Platform Evaluations & POCs
Enterprise Evaluation

Enterprise Agent Runtime Evaluation

Assessed runtime behavior, observability maturity, and gateway integration across enterprise agent platforms.

Agent RuntimesObservability

Assessed runtime behavior, observability maturity, memory handling, and gateway/API integration across enterprise agent platforms. Built working agents to validate flows & limitations; produced a leadership-level POV artifact.

Agent RuntimesObservabilityMemoryGateway
Proof of Concept

MCP (Model Context Protocol) POC

Built MCP client/server prototype for tool-access patterns with Claude integration.

MCPTool UseClaude

Built MCP client/server prototype for tool-access patterns. Integrated agent with MCP for tool execution; validated tool-grounded responses and documented integration patterns for enterprise adoption.

MCPTool UseAgentClaude
Data Engineering & ML Foundation
2020

Customer Engagement Framework

Cloud-native PySpark pipeline for batch & streaming processing on AWS, Azure, and Cloudera.

PySparkAWSStreaming

Cloud-native PySpark pipeline for batch & streaming processing deployed on Cloudera, AWS & Azure for an enterprise healthcare customer engagement platform.

PySparkAWSAzureClouderaStreaming
2019

German Risk Credit Analysis

ML model achieving 89% accuracy using XGBoost, Random Forest & Logistic Regression.

XGBoostPython

ML model achieving 89% accuracy using XGBoost, Random Forest & Logistic Regression to predict credit risk from the German Credit dataset.

XGBoostRandom ForestPythonScikit-learn
2018

Loan Defaulter Prediction

Processed 8 lakh observations with 99% accuracy ML pipeline for credit risk decisions.

PythonML

Processed 8 lakh observations to build an ML pipeline predicting loan defaults with 99% accuracy, enabling better credit risk decisions.

PythonMLPandasFeature Engineering
06 / Open Source

GitHub Repos

Hands-on experiments and personal builds — from agentic AI workflows to full-stack apps.

RishieRich GitHub Avatar

RishieRich

AI Engineering · Agentic Systems · Evaluation Frameworks · Data Engineering

6+ Public Repos
3+ AI/GenAI Repos
View on GitHub

AI_101_Clinical_Data_POC

End-to-end LangChain + LangGraph RAG pipeline for clinical data — agentic workflows, structured output generation, and document Q&A on healthcare records.

LangGraphLangChainRAGClinical AI
Python 0

ARQ_Lang-Chain-Smith-Graph

Multi-agent orchestration experiments using LangChain, LangSmith observability tracing, and LangGraph state machine composition — building and inspecting agent flows end-to-end.

LangGraphLangSmithOrchestration
Python 0

ARQ_OpenClaw_ClawdBot_Experiment

Chatbot experiments combining OpenAI and Claude APIs — multi-model routing, conversational prompt flows, and tool integration patterns for agentic assistants.

OpenAIClaudeTool UseChatbot
Python 0

ARQ_Claude_Code_Test

Claude API integration experiments — AI-assisted code generation, tool use patterns, prompt engineering, and agentic development workflow testing with Claude Sonnet.

Claude APITool UseAI Dev
Python 0

ARQ_Code_Validator

Automated code QC validation framework — structural integrity checks, coding standards enforcement, and quality gate automation for data pipelines and AI-generated code.

QC AutomationValidationStandards
Python 0

ARQ_EatEasy_v1.0

Full-stack food delivery platform built with Python — end-to-end architecture, REST API design, data modelling, business logic, and system design applied in a real-world domain.

PythonREST APIFull-Stack
Python 0
07 / Certifications

Credentials

AWS Certified Solutions Architect – Associate (SAA-C03)

Amazon Web Services Jun 2021 — Jun 2028  ·  Active
ACTIVE

AWS Certified AI Practitioner (AIF-C01)

Amazon Web Services AI / GenAI Foundations
AI CERT

Databricks Certified Data Engineer Associate

Databricks Data Engineering

Databricks Certified Developer for Apache Spark

Databricks Apache Spark

Databricks Generative AI Fundamentals

Databricks Badge · GenAI

Databricks Lakehouse Fundamentals

Databricks Badge · Lakehouse Architecture

SAP LeanIX Practitioner Level 1

SAP LeanIX Enterprise Architecture

SAP LeanIX Practitioner Level 2

SAP LeanIX Enterprise Architecture
08 / Writing & Enablement

Published Insights

May 1, 2025 LinkedIn

The 2025 ELT Showdown: Snowflake vs. Databricks Unveiled

In-depth comparative analysis of Snowflake and Databricks — architecture, performance, and cost trade-offs.

An in-depth comparative analysis of two industry-leading cloud data platforms — Snowflake and Databricks — covering architecture, performance, and cost trade-offs.

3 Likes Read Article
Mar 11, 2025 LinkedIn

Revolutionizing AI: Beyond Models & Algorithms

How modern AI is transcending algorithmic roots to become a transformational force in enterprise systems.

Exploring how modern AI is transcending its algorithmic roots to become a transformational force in enterprise systems and decision-making frameworks.

3 Likes Read Article
Mar 11, 2025 LinkedIn

McDonald's Embraces AI: Revolutionizing Fast-Food Service

How McDonald's leverages AI to optimize operations, personalize experiences, and redefine fast-food service.

A real-world case study on how McDonald's is leveraging AI to optimize operations, personalize customer experiences, and redefine fast-food service at scale.

7 Likes Read Article
09 / Contact

Let's Connect

Open to discussing AI Architecture, Senior GenAI Engineering roles, and enterprise agentic AI and RAG systems initiatives. I am also happy to collaborate on data platform architecture, LLM evaluation frameworks, and AI operating models for regulated industries.

LinkedIn Rishikesh Pote
Email Send a message
Open to AI Architecture Roles

Rishikesh Pote

AI Engineering Architect & Senior GenAI Engineer

Pune, Maharashtra, India

LangGraphRAGLLM-as-Judge DatabricksAWSAgentic AI