William C.

Machine Learning and AI Engineer

Upwork rates: Pay band: $80 for a consultation session (two hours maximum) Hourly: $70-120 an hour Outcome based: $200/feature Need an ML system that actually works in production, not just a notebook demo? I build AI solutions that ship. My systems process 50,000+ requests daily with sub-200ms latency. I take projects from research prototype to deployed product, handling the messy middle that most ML engineers avoid. What I deliver for clients: ### Voice AI Agents I build voice AI agents and conversational pipelines that work the way real users actually talk, which is messy, interrupted, and unpredictable. I use Bland.ai to design and deploy production voice agents with reliable call handling, dynamic routing, and tool-calling backends that connect to your existing systems. At AskHumans, I built outbound and inbound call agents for lead qualification, appointment booking, and live customer support. I wired the agent into the client's stack via Calendly webhooks for live scheduling, then reviewed transcripts post-launch to find where calls were dropping or going sideways, retuning the prompt and flow logic until call satisfaction hit 96.5%. ### LLM Applications and RAG Systems I build retrieval-augmented generation pipelines, multi-agent systems, and LLM-powered tools using LangChain, LangGraph, and direct API integrations. Recent work includes a document processing system that cut manual review time by 70% and a synthetic data pipeline generating 100K+ training samples daily. ### Production Machine Learning Your model is only valuable if it runs reliably at scale. I handle the full stack: model optimization, API development with FastAPI, containerization with Docker and Kubernetes, and cloud deployment on AWS (SageMaker, Bedrock, Lambda, EC2). I have built systems that maintained 99.9% uptime while processing enterprise workloads. ### GPU Optimization and High-Performance Computing Slow inference kills user experience. I optimize models with CUDA, implement efficient batching strategies, and architect distributed computing solutions. Background includes work on HPC clusters at NIH and CMU research computing. ### Graph-Based ML and Knowledge Systems When your problem involves relationships and networks, I bring specialized experience in graph neural networks, knowledge graphs, and Neo4j. I have applied these techniques to drug discovery research, fraud detection, and recommendation systems. My background spans research and industry. Three years of ML research at NIH and Carnegie Mellon gave me strong fundamentals in experimental design and statistical rigor. Current work as an AI/ML engineer at a federal consulting firm taught me how to build systems that meet enterprise requirements for security, compliance, and scale. I communicate clearly and keep you updated without requiring you to manage me. You will get working code, documentation, and systems you can maintain after the engagement ends. If your project involves voice agents, large language models, machine learning infrastructure, or getting an AI system into production, send me a message with the details. I will tell you honestly whether I am the right fit. Looking forward to hearing from you. ### Skills section Voice AI & Conversational Agents: Bland.ai, webhook integrations (Calendly, CRMs), transcript analysis, call flow optimization LLM Research & Fine-Tuning: HuggingFace Transformers, domain-specific fine-tuning, RAG systems, AWS Bedrock, OpenAI & Anthropic APIs, vector databases (Pinecone, Weaviate) Multi-Agent Systems: LangChain, LangGraph, agentic NLP pipelines, graph-based coordination frameworks, multi-agent orchestration Machine Learning & Deep Learning: PyTorch, TensorFlow, Graph Neural Networks (GraphSAGE), GANs, VAEs, synthetic data generation, reinforcement learning, computer vision, statistical learning theory High-Performance Computing & GPU Optimization: CUDA programming, GPU optimization, GraphBLAS, distributed computing, HPC clusters (SLURM), parallel algorithms, multi-node training Cloud & MLOps: AWS (SageMaker, Lambda, S3, EC2, Bedrock, CloudWatch), Docker, Terraform, Kubernetes, CI/CD, infrastructure as code, model deployment Full-Stack Development: Python, JavaScript/TypeScript, React, Next.js, FastAPI, Flask, Node.js, REST APIs Data Engineering: PostgreSQL, MongoDB, Neo4j, Pinecone, Weaviate, ETL pipelines, data visualization Cheminformatics: RDKit, molecular property prediction, graph-based drug discovery, synthesis route planning, SMILES encoding, Tanimoto similarity, NetworkX Other: Secret clearance (TS/SCI in progress), NVIDIA Accelerated Computing Certified, C/C++, Git