Fractional AI and Analytics Leadership
Bowser Analytics builds and leads AI and advanced analytics capabilities for insurers, banks, and government agencies, where a model has to be explainable and defensible as well as accurate. Fractional leadership, interim coverage, and project delivery.
Experience. Innovation. Results. | Let’s solve what’s next.
Capabilities
Strategy and execution from the same person. I define the problem with your executives, design the approach, and stay to build it.
How I help organizations
They have a production problem, a governance problem, or a leadership problem, and the models are where it shows up.
Work stalls between the notebook and the decision. I build the pipeline layer, the deployment path, and the standards that make shipping routine instead of heroic.
Documentation is thin, validation is inconsistent, and nobody can explain why the model decided what it decided. I assess it honestly, tell you what an examiner will find, and fix it.
Someone left, the search is taking longer than expected, or the team has never had senior direction. I step in, hold the commitments, develop the people, and hand over a function that works.
Legacy infrastructure, legacy licensing, week-long processing cycles. I have moved these estates to Snowflake and the cloud with validated cutovers and measurable improvement.
CECL, IFRS 17, AML, fair lending, statutory reporting. I have delivered against all of them, and I know what the evidence file has to contain.
I will give you a straight assessment of where your capability actually stands and what the next twelve months should cost and produce.
Services
Ongoing leadership of your function at one to three days per week: roadmap, hiring, methodology, review cadence, stakeholder representation, and hands-on technical direction. Typically six to twelve months.
Full-time or four-day coverage for three to nine months through a search, a transition, or a critical program, so the function does not lose a quarter.
Where you stand across data, business intelligence, machine learning, automation, adoption, and privacy; what to build, what to buy, and what to stop. Two to four weeks.
Hire, structure, and develop the team; establish standards, documentation, and training; and transfer ownership so the capability outlives the engagement.
The pipeline layer that makes shipping repeatable: SageMaker, Airflow, MLflow, Git, environment promotion, model versioning, and production monitoring.
An independent review of your model estate against what an examiner will actually ask, with a prioritized remediation plan and delivery of the remediation if you want it. Four to eight weeks.
CECL, IFRS 17, AML and transaction monitoring, fair lending analysis, and recurring statutory reporting, delivered against a date that does not move.
Models placed inside the transaction: scoring APIs, inference endpoints with latency targets, and integration into quoting, origination, and servicing flows.
From problem definition to a production model your team can run and defend: data preparation, feature engineering, development, champion and challenger selection, validation, deployment, and monitoring.
LLM and retrieval augmented generation solutions over your operational data, taken to production, plus multi-model evaluation frameworks that compare commercial LLM families on your own data and prompts.
Snowflake, AWS, and Azure, with validated parallel runs, reduced runtime, and reduced licensing and infrastructure cost.
What your SAS estate costs, what it would take to modernize or exit it, and the execution: SAS to Viya, Snowflake, Altair SLC, Python, or PL/SQL, with code compatibility analysis and support enablement.
Industries
Engagements
Five shapes cover almost everything. The right one depends on how much of the problem is decision and how much is delivery.
| Shape | Typical length | When it fits |
|---|---|---|
| Orientation assessment | 2 to 4 weeks | You need an honest read on where the analytics capability stands before you commit budget. Fixed fee, defined deliverables. |
| Model governance review | 4 to 8 weeks | An examination, an audit finding, or a model estate that has to hold up. Deeper, evidence-led, with a prioritized remediation plan. |
| Project delivery | 3 to 6 months | A specific model program, platform build, migration, or regulatory obligation. |
| Fractional leadership | 6 to 12 months | One to three days per week. You need senior direction continuously, not a full-time executive. |
| Interim leadership | 3 to 9 months | Full-time or four days. A departure, a search, or a program that cannot slip. |
Selected results
Why work with me
Where a specialist is needed I bring one in and stay accountable for the outcome.
Inside a national carrier owning the rating and modeling system. Eight years on the vendor side at SAS Institute. Federal consulting delivery across seven agencies at Guidehouse. And as an owner since 2022.
CECL, IFRS 17, OCC model risk guidance, CFPB fair lending, AML and transaction monitoring, and state insurance filing. Governance is built in from the first sprint, not retrofitted before the exam.
I will run your function and set its standards, and I will also open the code, review the validation, and ship the pipeline.
Five of fourteen engagements since 2022 have been repeat or extended work. That is the number I would want to see if I were hiring a consultant.
Documentation, training, and knowledge transfer are part of the engagement, not an upsell. A good outcome is your team not needing me.
About
Jay Leatherman founded Bowser Analytics in 2022 after a career spent building analytics capabilities from four different vantage points: inside a national property and casualty carrier, on the vendor side at SAS Institute, in federal consulting at Guidehouse, and leading advanced analytics for insurers as an independent consultant.
He builds and leads data science functions, delivers models into production under regulatory constraint, and modernizes the platforms underneath them. His work includes CECL credit risk deployment, IFRS 17 delivery, OCC model risk validation, CFPB fair lending analysis, federal emergency program delivery, and underwriting and pricing modeling programs, alongside cloud migrations onto Snowflake and AWS.
Bowser Analytics is a member of the SAS Platinum Partner consortium and has served fourteen clients across the United States, Canada, Australia, and Mexico. Jay holds an M.A. in Applied Demography from Bowling Green State University and is an AWS Certified AI Practitioner. He mentors working professionals through MIT Professional Education’s No Code and Agentic AI certification program, and is based in the Greater Boston area.
Start a conversation
Most engagements begin with a two to four week orientation assessment that pays for itself in the decisions it prevents.
Tell me what is stuck and I will tell you honestly whether I am the right person for it.
