• Evaluated pilot competency to identify areas of focus in an industry leading flight safety initiative at 30 airlines.• Implemented evaluative competency model with learning scientists using observable behaviors and operational data.• Led team of 35 to develop a suite of applications, enabling exchange of flight and maintenance files with 150 airlines.• Reduced early career costs 70% by refocusing to serve cost-conscious flight schools and support at-risk pilots.• Built cloud infrastructure for learning data to evaluate pilot competency, increased teams’ adoption from 25% to 95%.• API platform 2X APIs and 3X usage with access patterns, lifecycle, and self-service automation built on Azure.

• Taught ML: DL (RNN, LSTM), NLP, TS (ARIMA, GARCH), MLC (RF, ensemble, class imbalance, PRC).• Taught financial technology in ‘Professional and Continuing Education’ as 2U bootcamp.

• Designed and coded models, mechanics, and strategy to participate in $16B FCC Auction 904.• Cluster analysis, network expansion scenarios with valuation, auction bids.• GIS data visualization of networks, census block attributes, and path obstructions.• Used GeoPandas, Shapely, Pygeos, Fiona, OSMNX, Numpy, SQL, Matplotlib, Google Earth, Sklearn, Scipy.Consultant for a short-term project.

• Mobile app for largest European electric car charging network as part of Fortum Digital Mobility.• Created a subscription feature, lowering marginal charge fee, increasing purchase conversion rate from 11% to 17%.• Innovated a transaction reconciliation workflow that doubled profitability margin from 4% to 8%.• Decomposed a monolith into micro-services to scale infrastructure during rapid growth.• Integrated route planning using EV model, charge speed, availability, amenities, configuration and probabilities.• Applied ML to identify overdue invoices likely to be paid, recovered €1M outstanding when reissued.

• Refined a website builder to be intuitive and visual, increasing site launch rate from 20% to 90%.• Improved internal search with Elasticsearch implementation for 40k employees.Employed through a 3rd-party.

• Built a mobile app using novel loans (anonymous, efficient) to solve the friends-and-family lending problem.• Disintermediated lending using a social graph with 1-hop adjacency lists and multi-hop funding."You know credible people"

• Designed UI to make complex-options trading as easy as trading stock with my IP (patent WO2013166415A1).• Rule-based expert system augmented with ML for complex options strategy composition and selection.• Feature engineering approach where each strategy has a weighted geometric structure related to an asset.• SMEs generated labelled data by selecting strategies in simulation for validation.• Model tuned with human-in-the-loop (HITL).• Attracted 500 B2C users and 4 contracts for B2B2C distribution in support of a B2B growth strategy.• Ran a business (venture capital, hiring, strategy, Agile, roadmap, vision, prioritized features, market research).Orca was a rule-based expert system augmented with ML techniques for complex options strategy composition and selection.It used a feature engineering approach where each strategy has a weighted geometric structure and relationship to an underlying asset. SMEs labelled data by selecting strategies in simulation. Market data snapshots are unpredictable and option chains (strikes, expirations, liquidity) are sparse, so labelling was for validation, not training. SMEs reinforced confidence in my strategy composition and financial engineering as I observed high precision against labelled data.To improve the model it traversed underlying forecast space bins (3 axis: time, price, implied volatility) to generate ranked output from millions of market snapshots. After every epoch, data visualizations with frequency statistics allowed human-in-the-loop evaluation, while the system tuned parameters.In production, a simple UI let users input their underlying forecast (the same info used to trade stock) and it presented 5 complex options strategies. Each strategy was distinct, and expertly composed for the underlying forecast. Users could quickly and confidently trade. Orca leveled the playing field, protecting people from professionals. Wall Street traders are “sharks”; killer whales are apex predators.

• Improved an AI chatbot to influence behavior, prompting $4B (4X YoY) asset inflow with 7M users.• Decision trees were used as a foundation for user targeted messaging.• A/B tests layered on a decision tree framework for data-driven optimizations, updating branches and thresholds.• New variants evolved for segments defined by web activity data, demographic data, and account activity data.

• Helped the most sophisticated clients formulate financial derivative strategies for market scenarios.I was recognized consistently as a "High Performer" among derivatives brokers as a member of the thinkorswim team.

Provided support to self-directed investors during The Great Recession. Consistent "High Performer" (top percentiles).