// about
Charles Wang
Senior Strategic Projects Lead, Gen AI · Scale AI
Engineer and investor focused on AI systems and data infrastructure. At Scale AI I work on frontier post-training data for agent environments and coding agents, partnering with top research labs on model development. Previously principal investor at Good AI Capital, backing early-stage data and AI companies. Before that, four years at Palantir building and leading deployed engineering teams on data products and predictive ML systems. B.S.E. Computer Science (summa cum laude), University of Michigan. M.B.A., University of Chicago Booth.
// experience
Leading strategic projects in the New Product and Frontier Data group for Agent Environments and Coding Agents. Delivering frontier post-training data to top research labs, driving model development and generating millions in revenue.
Focused on data & AI infrastructure and enterprise investments. Responsible for deal sourcing, due diligence, portfolio management, and industry research. Sourced a Seed deal in Coblocks and supported a Series A deal in Persist AI.
Supported three investments in Modal Labs (Series A), Databricks (Series I), and PuppyGraph (Pre-Seed).
Sourced and evaluated OSS startups; generated a $5.5M Seed-stage deal in Wherobots (Wing VC co-invest).
Led a deployed team of six at a railway client building locomotive monitoring software. Built time series regression models for predictive maintenance, capturing $20M+ in annual cost savings. Drove 5× user growth on a high-visibility cybersecurity product.
Identified mobile ad viewability issues within YouTube and third-party Google ads. Released an improved occlusion detection algorithm to increase mobile ad viewability reporting accuracy by 2%.
Built infrastructure to replicate customers' mobile ad conversion tracking ecosystem on AdWords and AdMob networks.
// investments
// writing
// projects
Genome indexing tool to detect DNA mutation sequences across thousands of cancer patient records. Poster.
Sentiment analysis system pooling Twitter data to generate unbiased movie ratings. Paper · Poster.
8×8×8 LED cube with an FPGA programmed in assembly to visualize two-variable functions.
Automated vinyl record cleaning machine built as a senior project (team of two).
Built a portable MP3 player for Stanford's summer engineering design course (individual).