Lead AI Solutions Architect — Generative AI · Agentic Systems · RAG · LLM Platforms UK-based · boparaji30@gmail.com
I design and productionize scalable ML and Generative AI systems on AWS, Azure, and Databricks. Nine-plus years turning messy, unstructured data into secure, governed, enterprise-grade AI products — with deep focus on RAG, multi-agent orchestration, LLMOps, and foundation-model integration.
Trusted advisor to business stakeholders — bridging AI strategy and delivery with reusable services and domain-aligned AI capabilities that balance governance, security, cost, and speed to value.
- [Alignerr / Labelbox] — RL Task Author & AI Training Expert. Authoring deterministic, code-graded reinforcement learning tasks for frontier LLM labs; designing multi-metric rubrics with calibrated credit curves; running Monte Carlo validation across skilled-agent and baseline distributions.
- [Micro1] — AI Trainer. Prompt engineering, evaluation-framework design, and human-in-the-loop validation supporting SFT and RLHF pipelines across coding, reasoning, and business domains.
- Open to my next senior GenAI / AI-platform engagement — most recently architected an end-to-end AWS AI platform at Magnus Consulting (LLM schema mapping → S3/Athena → Pinecone → production RAG → multi-agent GTM report generation, fronted by a Next.js chat app on Amplify + EC2).
| Role | Company | Period | Focus |
|---|---|---|---|
| Sr AI & Data Platform Engineer | Magnus Consulting | Aug 2025 – Feb 2026 | GenAI platform, multi-agent RAG, LLM schema mapping |
| Lead Data Scientist (GenAI) | Unisys Logistics | Nov 2023 – Aug 2025 | Multi-agent LangChain workflows, LLM router, MCP tool server |
| ML Design Lead | easyJet | Apr 2023 – Oct 2023 | OCR at 2M docs/yr, Bedrock/OpenAI RAG (+40% relevance), prompt-to-video (–70% production time) |
| Senior Data Scientist | ACAS | Oct 2022 – Apr 2023 | NLP topic modelling and multi-class classification for national helpline transcripts |
| Senior Data Scientist | Salesforce | May 2021 – Sep 2022 | Forecasting, Neo4j customer knowledge graph, 5TB/day BigQuery pipelines |
| Risk & Uncertainty Consultant | AstraZeneca | Jul 2019 – Feb 2021 | UQ tooling for GCA risk assessment; COVID-19 studies |
| ML Engineer | National Nuclear Laboratory | Jan 2017 – Jun 2019 | Published neural surrogate models for chemical-plant sensitivity analysis |
Generative AI Architecture — RAG, foundation-model integration (OpenAI, Llama, Bedrock), vector search (FAISS, Pinecone, Databricks Vector Search), multi-agent orchestration (LangGraph, CrewAI, OpenAI Agent SDK), MCP integrations, agentic workflow design. AI Platform / Cloud — AWS (S3, Redshift, EMR, Kinesis, SageMaker, Bedrock, Lambda, Amplify, EC2), Azure, Databricks, private-endpoint model access, Terraform IaC, Docker / Kubernetes / Helm. LLMOps / MLOps / AgentOps — CI/CD for models and services, automated eval + benchmarking, drift detection, MLflow, real-time inference APIs. AI Security & Governance — IAM least-privilege model access, PII masking in prompts, secure egress/ingress, responsible AI evaluation. Advanced ML — Bayesian modelling, forecasting, optimisation under uncertainty (stochastic + MIP), NLP, computer vision, deep learning.
- Probability-Bound-Analysis — Python library for constructing confidence boxes under epistemic uncertainty. Rooted in my PhD.
- PyIPM — Python port of the OpenCossan Interval Predictor Model toolbox.
- LSTM — Recurrent-net implementations from earlier deep-learning work.
- Cheque — Remote check deposit via image recognition.
- PhD, Engineering — University of Liverpool
- PhD, Engineering — National Tsing Hua University, Taiwan
- MSc Energy Generation
- BEng Aerospace Engineering — University of Liverpool
- Interests: LLMs, NLP, Bayesian inference, Monte Carlo simulation, algorithms.
Open to advisory work and interesting problems in RAG, agentic systems, LLM platform engineering, and AI training / evaluation. Best reached at boparaji30@gmail.com.