Vetted Latency Optimization Professionals

“At PwC, built and productionized an agentic RAG enterprise search assistant over 6M internal documents (8M embeddings), deployed across AWS and GCP. Drove major retrieval gains (72%→92% precision via BM25+dense hybrid with RRF and cross-encoder re-ranking), reduced hallucinations 30%, achieved <2s latency at 50–60K queries/month, and cut support tickets 30%—boosting adoption to 2,500 users by adding source-cited answers.”

“Hands-on LLM practitioner who built a production document-processing pipeline in Python, tackling long-document handling and latency with chunking/batching and a user-driven correction feedback loop. Experienced operationalizing AI workflows with Kubernetes (CronJobs, autoscaling, scheduled data cleaning and weekly retraining) and applying structured testing/evaluation (E2E, LLM-as-judge, HITL) while communicating solutions clearly to non-technical clients using visual diagrams.”

“Systems-focused engineer who built an OS kernel with multithreading, priority scheduling, system calls, and synchronization primitives, and debugged race conditions end-to-end. While not yet hands-on with ROS/SLAM, they clearly connect low-level concurrency and scheduling decisions to deterministic, reliable robotics-style real-time workloads.”

“Built and productionized LLM-driven sentiment scoring for earnings call transcripts at Goldman Sachs, replacing legacy NLP to deliver a cleaner trading signal while managing latency/cost via batching, caching, and distilled models. Also implemented an Airflow-orchestrated fraud modeling pipeline at MetLife with drift-based retraining and SageMaker deployment, and has a disciplined evaluation/rollout framework for reliable AI workflows.”

“Built a production LLM-driven personalization microservice for realtor.com using LangChain/LangSmith with an MCP tool layer and RAG to generate schema-constrained ranked listings in real time, replacing a rule-based engine and improving engagement/lead conversion. Also owned an ambiguous cross-channel identity initiative, implementing an identity graph via Twilio Unify with required SDK and data-warehouse integration.”

“LLM/agentic systems practitioner who specializes in moving demo-only assistants into reliable, observable, cost-controlled production services. Strong in real-time diagnosis of complex agent workflows (including tracing, loop detection, and guardrails) and in customer-facing enablement—running workshops, building tailored PoCs, and partnering with sales to close deals by proving reliability in high-risk pilots.”

“Embedded C++ engineer with Bosch automotive infotainment experience, owning real-time audio middleware modules with strict latency/memory constraints. Strong in profiling/optimizing deterministic behavior, debugging hardware-specific intermittent issues, and building automated test + CI pipelines; currently ramping up on ROS2 concepts (DDS, nodes/topics/services) to transition toward robotics.”