Carefull is an AI-powered financial safety platform that helps banks, credit unions, and wealth advisors protect older-adult customers from fraud and money mistakes. We help financial institutions maintain whole-family relationships while protecting their clients. Carefull’s technology addresses senior-specific financial safety challenges: our monitoring detects fraud patterns missed by industry-standard tools, and our features — identity-theft protection, password and document management, communication tools, and how-to content — help customers maintain financial independence while enabling loved ones to step in when needed.
The RoleWe are looking for a Senior AI Engineer to join our Data team and build, evaluate, and improve the AI-powered systems at the core of our product. A big part of the work is detection: systems that analyze financial transactions and decide whether to alert a family that something concerning may be happening with their loved one's money. You'll also dig into user behavior and patterns, including how people interact with the product, and use what you learn to shape what we build next. This is a hands-on role. You'll research fraud patterns, design detection logic, write production code, and rigorously evaluate system performance. You'll own features end to end: from understanding a problem, to implementing and deploying a solution, to measuring whether it actually works.
How We WorkOwnership here means caring about the outcome, not only the delivery. We work in small, fast increments, because a focused change in front of users today teaches us more than a complete one next week. You'll have a lot of autonomy in how you approach problems, and we trust people to find the next step on their own. When you're stuck, a quick question is always welcome, and short, frequent updates go a long way on a remote team. We use AI coding tools heavily and expect you to as well. We also expect you to understand what you ship and to be able to explain the reasoning behind every change.
What You’ll DoDesign and ship new AI-driven detection features, from first prototype to production.
Build data enrichment pipelines that extract structured information from messy, real-world financial transaction data.
Research fraud and scam typologies relevant to older adults, and translate that understanding into detection logic that works at scale.
Build reproducible evaluations (test sets, metrics, error analysis) so every change can be measured against the last one.
Investigate issues reported by users or surfaced in production, find the root cause quickly, and ship the fix.
Optimize AI pipelines for accuracy, latency, and cost, making informed tradeoffs about model selection and system architecture.
Work with Customer Care, Go-to-Market, and partner-facing teams to understand what real users need.
Keep up with new developments in LLMs and agents, and find practical ways to use them here.
Required
Strong Python skills, with experience building data pipelines and production systems.
Hands-on experience building LLM applications in production: prompting, structured outputs, context management, and working directly with provider SDKs and APIs.
Comfort deploying and operating what you build on a cloud platform.
A habit of measuring before claiming something works. You know how to set up an evaluation, read precision and recall, and dig into errors until you understand them.
A track record of owning work end to end without close supervision.
Real curiosity about the domain. You'll want to understand how the US financial system works, how money moves between accounts, and how scammers take advantage of it.
Comfort reasoning about ambiguity. Our domain is full of cases where the answer depends on context, and you need to build systems that handle that.
Clear written and verbal communication in English. You'll document your reasoning, present to stakeholders, and explain technical decisions to non-technical teammates.
Strong Plus
AWS experience (Lambda, CDK, Bedrock, Redshift, DynamoDB).
Experience with LLM observability and tracing tools such as Langfuse or LangSmith.
Background in fraud detection, fintech, or risk and compliance.
Experience with financial transaction data (ACH, Zelle, wires, card payments).
Experience working with regulated institutions such as banks
Nice to Have
Experience working with regulated industries or bank partners.
Exposure to elder care, aging-in-place, or financial vulnerability research.
Background in data science or ML beyond LLMs (statistical modeling, anomaly detection).
Silver Screening interview
Take-home challenge
Client technical interview
CTO interview
Final interview Hiring Manager
Skills Required
- Strong Python skills and experience building data pipelines and production systems
- Hands-on experience building production LLM applications, including prompting, structured outputs, context management, and provider SDKs or APIs
- Experience deploying and operating systems on a cloud platform
- Ability to design evaluations, use precision and recall, and perform error analysis
- Track record of owning work end to end without close supervision
- Curiosity about the US financial system, money movement, and fraud methods
- Ability to reason about ambiguity and build context-aware systems
- Clear written and verbal communication in English
- AWS experience with Lambda, CDK, Bedrock, Redshift, or DynamoDB
- Experience with LLM observability and tracing tools such as Langfuse or LangSmith
- Background in fraud detection, fintech, or risk and compliance
- Experience with financial transaction data such as ACH, Zelle, wires, or card payments
- Experience working with regulated institutions such as banks
- Experience with regulated industries or bank partners
- Exposure to elder care, aging-in-place, or financial vulnerability research
- Background in data science or machine learning beyond LLMs, including statistical modeling or anomaly detection
What We Do
Silver.dev is a technology company focused on developing, maintaining, and distributing mobile applications as its core product, tackling deep tech challenges.
