Felix Elias

Legal Technology Center of Excellence

Founded and led a Legal Technology Center of Excellence at Accenture, significantly enhancing team capabilities and market presence.

• Founded and grew a Legal Technology Center of Excellence to 200+ trained practitioners within the first six months of launch
• Architected the "AI Legal Front Door" offering — a multi-agent system integrating 7 CLM platforms — generating $1M+ in new sales pipeline
• Established strategic vendor relationships across CLM, eSignature, workflow automation, and AI domains
• Led RFP responses and go-to-market positioning that directly secured multiple client engagements

Generative AI Risk and Compliance

Led Gen AI implementation for a large global bank to perform control to regulation mapping and assess control coverage.

• Designed and implemented a GPT-4-based control-to-regulation mapping solution for a large global bank
• Applied 5W1H methodology to assess control completeness across a regulatory control library
• Used generative AI to surface and rewrite deficient control elements, reducing manual review cycles
• Delivered a production compliance automation solution supporting MRA-driven remediation workflows

Legal Transformation Program

Led multiple work streams for a large global capital markets firm across Legal Hold, Litigation Operations, Board Management, and Outside Counsel Management.

• Led multiple concurrent workstreams for a global capital markets firm spanning Legal Hold, Litigation Operations, Board Management, and Outside Counsel Management
• Designed a comprehensive legal transformation roadmap adopted as the foundation for a broader organizational initiative
• Implemented collaborative workflow solutions for outside counsel matter management and spend visibility
• Drove engagement expansion from initial assessment to full-scale legal transformation program

Intelligent Automation Platform Development

Initiated development of comprehensive platform for global bank to streamline regulatory contract remediation process.

• Architected and delivered an end-to-end intelligent automation platform for a global bank in four months, supporting regulatory contract remediation
• Integrated document management, Blue Prism RPA, workflow automation, and business intelligence layers into a unified platform
• Realized efficiency gains equivalent to 2+ FTEs and reduced manual processing time by up to 80% across targeted workflows
• Directed an RPA Automation Factory across Blue Prism and UiPath engagements, at peak managing automation delivery for up to 7 FTE reductions

Enterprise Automation Assessment

Conducted comprehensive assessment of nearly 500 applications at tier-1 global bank to identify automation opportunities.

• Conducted a comprehensive automation opportunity assessment across nearly 500 enterprise applications at a tier-1 global bank
• Evaluated candidates against Intelligent Automation, RPA, NLP, NLG, and Business Intelligence frameworks
• Produced a detailed technology investment roadmap with prioritized implementation sequencing
• Assessment findings directly shaped the bank's multi-year automation strategy and vendor selection

Innovation & Emerging Technology

Developed cutting-edge solutions including VR transaction monitoring and privacy lifecycle management platforms.

• Developed a Virtual Reality transaction monitoring proof of concept, demonstrating compliance workflows inside a VR environment — presented during client innovation campaigns
• Built a Privacy Lifecycle Management solution on the Appian platform, including regulatory tracking and metrics dashboarding
• Led a Federal Reserve MRA-driven KYC compliance program for a large financial institution, coordinating remediation across multiple control domains
• Applied emerging AI capabilities to legal, compliance, and financial services contexts across engagements — from POC through production

Production-grade agentic AI application | LangGraph · FastAPI · Gradio · MCP

A fully production-grade agentic AI system built as the capstone for the Ready Tensor Agentic AI certification. The system uses a multi-agent LangGraph architecture to automatically evaluate programming assignments end-to-end — from code parsing through feedback generation.

• Multi-agent LangGraph pipeline with specialized agents for code analysis, rubric evaluation, and narrative feedback generation
• Custom MCP server with 30+ tools exposing the grading system's capabilities to AI clients
• RAG pipeline with vector embeddings for rubric retrieval and contextual evaluation
• Multi-LLM support (OpenAI, Anthropic, open-weight models) with LangSmith observability
• FastAPI backend, Gradio frontend, 130+ automated tests with CI/CD via GitHub Actions
• Presented as a 15-slide technical capstone demonstrating end-to-end agentic system design

• LangGraph
• FastAPI
• Gradio
• MCP
• RAG
• LangSmith
• Python
• CI/CD
• Multi-Agent

PEFT fine-tuning on LLaMA 7B | LoRA · QLoRA · Axolotl · DeepSpeed

A personal research and engineering project for fine-tuning open-weight large language models using parameter-efficient methods. Built to develop hands-on expertise in the full fine-tuning stack — from dataset preparation through model evaluation.

• Fine-tuned LLaMA 7B using LoRA and QLoRA (4-bit quantization) via Axolotl and DeepSpeed
• Implemented PEFT training workflows with Hugging Face Transformers, TRL, and Accelerate
• Experimented with instruction-tuning datasets and custom prompt templates
• Applied Weights & Biases for training observability and hyperparameter tracking
• Developed practical expertise in quantization tradeoffs, adapter merging, and inference optimization

• LoRA
• QLoRA
• Axolotl
• DeepSpeed
• Hugging Face
• PEFT
• LLaMA
• PyTorch
• W&B

Model Context Protocol server | HTTP transport · Vercel deployment

A production-deployed MCP (Model Context Protocol) server built to expose structured tools to AI agents and LLM clients. Deployed via HTTP transport on Vercel, making agentic tool calls available to any MCP-compatible client.

• Designed and implemented a full MCP server following the Model Context Protocol specification
• Deployed on Vercel with HTTP transport — accessible to any MCP-compatible AI client
• Exposes structured tools enabling AI agents to interact with external data and workflows
• Applied learnings directly to larger agentic projects including the Intelligent Assignment Grading System

• MCP
• Model Context Protocol
• Vercel
• HTTP
• Agentic AI
• Python
• API Design

Agentic AI workflow | LangGraph · FastAPI · Next.js · Railway

An independent pipeline project applying agentic AI to an everyday challenge: intelligent job search. The system uses a multi-agent LangGraph backend to automate job discovery, relevance scoring, resume tailoring signals, and application tracking — with a Next.js frontend and FastAPI API layer.

• Multi-agent LangGraph architecture with specialized agents for search, scoring, and content generation
• FastAPI backend deployed on Railway; Next.js frontend deployed on Vercel
• Demonstrates applied agentic AI engineering across the full stack — from orchestration to UI
• Active development project reflecting current work in production-grade agentic pipelines

• LangGraph
• FastAPI
• Next.js
• Vercel
• Railway
• Multi-Agent
• Python
• TypeScript