Vetted Research Assistants in the Bay Area

“LLM/agent engineer who built a production code-generation agent at Corvic AI that lets non-technical users query CSV/tabular data in natural language by generating and executing Python. Focused on making LLM systems reliable and scalable via schema-aware validation, sandboxed execution-feedback retries, prompt caching/embeddings, async execution, and high-throughput data processing with Polars; also partnered with Adobe product/marketing to ship brand-aligned AI content generation for email and push notifications.”

“ML/NLP engineer with experience at NVIDIA and Cruise building production-grade AI systems across genomics/biomedical research and autonomous vehicle data. Has delivered multimodal LLM pipelines, large-scale entity resolution, and hybrid semantic search (BERT embeddings + FAISS + Elasticsearch), with measurable impact (≈40% accuracy/retrieval gains; ≈30% data consistency improvement) and strong MLOps practices (Kubernetes, CI/CD, MLflow, Prometheus/Grafana).”

“Embedded/robotics software engineer with Hyundai Motors experience who owned an AI-driven perception validation pipeline using a Transformer-based approach to generate stable synthetic in-cabin audio for autonomy/ASR testing, cutting downstream testing time by 50%+. Has hands-on ROS integration (IMU sensor streaming, inference, control nodes), MQTT-based distributed messaging, and cloud/container deployment experience (Docker, Node/Express, AWS, CI/CD).”

“Built and shipped a production LLM-powered medical scribe that generates structured clinical visit summaries using RAG, strict JSON schemas, and post-generation validation to reduce hallucinations. Experienced in making LLM workflows deterministic and observable (structured logging/metrics/tracing) and in evaluation-driven iteration with metrics like schema pass rate and edit rate; collaborated closely with clinicians and policy stakeholders at Scale AI to drive adoption.”

“Robotics-focused engineer with ROS 2 experience who has built and debugged real-time, distributed control/orchestration systems under production-like latency and safety constraints. Led platform changes at Persona for a real-time verification orchestration system using deterministic state machines and async workers, and has hands-on experience stabilizing multi-robot navigation/SLAM behavior using rosbag, RViz, and stress testing in simulation (Gazebo).”

“Robotics software engineer (master’s student) who placed 3rd in the CMU VLA challenge and presented at IROS, building an LLM-powered language system (Gemini 2.5) for mobile-robot scene Q&A and language-based navigation. Hands-on ROS1/ROS2 experience including ros2_control + PILZ planning for a KUKA arm, plus simulation (Gazebo) and containerized submissions with Docker.”

“Robotics/embedded builder who trained an aviation-specific LLM and deployed it offline on an NVIDIA Jetson for an in-flight voice assistant, solving performance and cabling constraints with NVMe storage and Bluetooth. Also has hands-on Raspberry Pi/Arduino robot builds (including a cigarette-butt picking prototype with hydraulic actuation) plus Docker-based FEA work using FEniCS/Gmsh and strong CI/CD + automated testing practices.”

“Product/UX designer who builds end-to-end systems across both consumer wellness and industrial/technical domains. Designed BloomPath (mental-wellness platform for therapists and young professionals) using research-driven, emotionally safe interaction patterns, and also simplified a Bosch autonomous parking vision-language mapping pipeline into a developer-facing real-time UI with layered debug tooling. Comfortable collaborating deeply with engineers and contributing in React/JS.”

“Spatial computing software engineer focused on making Apple Vision Pro/visionOS accessible, including building VoiceOver and Live Captions features. Debugged a complex Live Captions issue involving dual audio inputs during FaceTime screen sharing by leveraging iOS implementation docs and creating concurrent audio sources; also has safety-critical testing experience from train control systems and is interested in pivoting into robotics.”

“Built and productionized a dual-agent LLM invoice-processing system for GFI Partners, adding guardrails and audit trails to earn stakeholder trust and drive adoption while cutting operational burden by 75%. Uses LangSmith observability to diagnose real-time workflow regressions and has experience teaching agentic AI concepts (e.g., at Carnegie Mellon) through hands-on, scaffolded demos.”