
Chemistry evaluation, frontier model-training support, and expert prompt engineering.
Selected work
The common thread is not an industry or a deliverable. It is turning difficult chemical judgment into something other people can inspect, trust, and keep using.
01 Chemistry software
02 Scientific data quality
03 AI evaluation
04 Workflow architecture
01 / FLAGSHIP
NewChem™ Database 0.67k
NewChem™ began as a molecular database and grew into ten connected desktop tools plus a packaged CLI for local chemistry automation.

10working tabs in one desktop app
0preloaded molecule records
1local library under your control
02 / SELECTED COLLABORATIONS
Experience behind NewChem™

Chemistry evaluation, frontier model-training support, and expert prompt engineering.

Sponsor of the public series Important Papers in Organic Synthesis.
03 / PRACTICE
Selected areas of work
PhD-level organic chemistry prompts, rubrics, failure taxonomies, and review workflows designed to distinguish chemical reasoning from confident imitation.
Chemical validity, ambiguous records, annotation logic, lineage, and review gates examined as one data-quality system.
Independent scrutiny of routes, mechanisms, method choices, and experimental assumptions when the cost of a weak answer is real.
Expert judgment translated into interfaces, schemas, decision logic, and handoff artifacts that technical teams can operate.
Client and collaborator work is described by problem class. Confidential names, data, and implementation details stay confidential.
04 / DIRECTION
The NewChem™ roadmap
NewChem™ expands after the existing workflow is dependable. Reliability, cross-platform packaging, clear documentation, and user feedback come before feature theatre.

The next useful thing
If your chemistry lives in one expert's head, a fragile spreadsheet, or a system nobody trusts, there is work worth doing.