Cesar Salcedo Email and Phone Number

- Led the development of algorithms for automatic data collection from the web, building a comprehensive company database from web pages, documents, and videos using LLMs (OpenAI, Anthropic, Cohere, Llama.v2, LlamaIndex, LangChain).- Developed full-stack web applications with TypeScript, Next.js, React, Tailwind, and Shadcn.- Implemented web scrapers using BeautifulSoup, Selenium, and Docker.- Managed databases (Postgres, MySQL, Prisma, Supabase, Pinecone, Chroma).- Deployed services via Vercel and Render.- Ensured code quality with testing suites (Pytest, Django-Pytest, Unittest).

- Studied zero-shot RL and option discovery via sequence modeling for hierarchical, goal-conditioned RL.- Implemented exploration algorithms (PSRL, UCRL), RL environments, and transformer models with PyTorch and HuggingFace.

- Received one-to-one mentoring from Google researchers- Contributed to open source project for machine learning audio research- Panelist at Alumni Preso Week 2023b

- Contributed as research engineer for implicit interactive fleet learning project with AutoLab.- Implemented behavioral cloning algorithms (Dagger, DART).- Integrated IsaacGym and custom Gym environment benchmarks into codebase.- Established experiment execution pipeline to ensure reproducibility and comparable results.- Automated parallel and distributed execution of experiments across GPUs and servers via W&B and shell scripts (+8x speedup).

- Author of two papers on multi-agent deep reinforcement learning.- Implemented and vectorized multi-agent control algorithms on RLLib, TensorFlow and NumPy (2x speedup).- Designed custom multi-agent RL environments with Gym.- Contrasted rule-based and deep RL algorithms empirically across cooperation tasks.

- Conducted project on unsupervised image anomaly detection via likelihood estimation from generative models- Implemented GANs, VAEs, and normalizing flows in PyTorch.- Improved out-of-distribution generalization with pre-trained feature extractors (ResNet, SimCLR, CLIP).- Wrote script for automatic execution of runs for hyperparameter tuning using Screen and bash scripts (3x speedup).

- Preprocessed 10 datasets into Wolfram Language data structures.- Added 8 API integrations to Mathematica.- Automated API integration pipeline.