Senior Member of Technical Staff (Applied AI)

#

Senior Member of Technical Staff (Applied AI)

Location: New York City

Why You Should Join Us

Solstice is redefining how life sciences organizations commercialize their therapeutics. We are building a commercial engine that allows pharmaceutical marketers to launch campaigns at 100x the speed.
 
- Rapid growth: Over the past year, we have been working with some of the top life sciences manufacturers and are working with over 50+ pharma brands.
- A hard technical problem: Every claim in a piece of pharma marketing content has to trace back to an approved source. Fair balance rules govern what appears next to it. A brand's medical, legal and regulatory reviewers see everything before it ships, which is why a single campaign can take months to reach a physician. Building a system that produces content which is both persuasive and defensible turns out to be a genuinely hard problem in constrained generation, and there's no playbook for it. We have the customers and the proprietary data to work on it properly.
- Top-tier investors: We've raised from investors like Transformation Capital, Twelve Below, Virtue and the founders of Datavant, Commure, and Paradigm to supercharge our growth and build an elite team of engineers and operators.
 
This is a senior role on a small team. Whatever architecture you land on in your first six months is what we'll be living with for years. We're looking for someone who wants that kind of responsibility rather than someone who needs to be handed it.
   
## About the Role

We're looking for a senior Applied AI engineer to own the technical direction of our AI systems: retrieval, generation, verification, evaluation, and everything involved in keeping all of it running in production. The architecture here isn't settled, and we're not hiring someone to build what's already been decided. When there's a real fork in the road, you're the one who picks. We'd honestly rather hire someone who argues with how we've done things so far. Beyond the systems themselves, you'll work closely with our VP of Engineering, Jay, as well as co-founders Aris and Yiwen on product strategy, technical hiring, and how we position ourselves in a market that moves fast.
 
### What You'll Be Working On

- Output a brand team is proud to ship: Our clients demand craft, not AI-slop. Compliance is the floor here, not the bar. Pharma assets are visual, things like emails, banners, IVA slides and leave-behinds, so fair balance is as much a layout problem as a copy problem. ISI has to carry comparable prominence. Brand systems have to hold up. Hierarchy and typography need to survive generation intact. Rejected claims aren't really our worry. What keeps us up is an asset that clears review and still looks machine-made, so the marketer quietly rewrites it by hand. Getting a model to produce work with real craft is harder than it sounds and mostly unsolved.
- Tying content back to prescriptions delivered: Getting through MLR review means an asset can ship. It doesn't mean it did anything. Over time we want to connect what we generate to how it performs in market, including engagement, HCP response and eventually scripts written, then feed that back into generation so the system learns what works and not just what passes. That means building the measurement layer, working out what can honestly be attributed to the content itself versus everything else shaping a prescribing decision, and turning it into a real iteration loop. Nobody has done this well yet. If we get it right, what we're optimizing against stops being a click rate and becomes a prescription filled, and a patient starting therapy sooner.
- Evaluation when ground truth is slow and expensive: There's no cheap way to answer "is this compliant?" The real signal comes back weeks later as reviewer redlines, written by people for other people, with no consistent format and no agreement between brands about exactly where the line sits. Getting to an offline metric that tracks downstream MLR pass rate is the thing most of our other work depends on.
- Claim-level attribution: Grounding in this context doesn't mean the model saw a relevant document. It means a given sentence can be defended against a specific source, and a reviewer can be shown which one. Our corpora are messy and proprietary, covering prescribing information, clinical data, previously approved assets, brand style guides and competitive claim libraries, and provenance has to survive all the way through generation into the finished asset.
- ML Ops and behavioral regression: The usual pieces, meaning CI/CD for ML, model versioning, automated testing and monitoring, plus the harder part underneath them. Model upgrades shift output distributions in ways that averages hide, and in this domain the tail is where all the risk lives. Catching that before a client does is ongoing work rather than a one-time setup.
   
## Some Ideal Traits We're Looking For

- Production ownership: You've built an ML or LLM system, shipped it, watched it degrade in ways you didn't see coming, and fixed it. Prototypes and demos aren't the experience we're hiring for.
- Evaluation instinct: You've had to invent the measurement before you could improve the system. This is the strongest single signal for us, so lead with it if it describes you.
- Depth in modern LLM systems: Retrieval architectures, agentic pipelines, fine-tuning when it earns its place, and a clear sense of when each one is the wrong tool.
- Strong proficiency in full-stack web development technologies such as Python, FastAPI, Docker, AWS. You can take something from research question to deployed service without handing it off.
- Judgment under ambiguity: A lot of this has no established answer yet. We need someone who can commit to a direction on incomplete information, be wrong quickly, and say so.
- Superb communication and collaboration skills: We believe in direct communication and willingness to provide as well as accept criticism.
     
## Benefits

- Health, dental, and vision insurance
- Ground-floor equity opportunity
- Visa sponsorship (O-1, H-1B, TN) available
- Unlimited PTO
- 401(k) + 4% company match
- In-office lunches 5 days/week
- Professional growth stipend
- Relocation support
   
## Competitive NYC Compensation

This role offers a highly competitive NYC salary in the $230,000 to $350,000 range, calibrated to reflect experience, seniority, and the level of ownership you'll take on as part of the early team. We benchmark against top-tier startups and established tech companies to ensure our compensation is both fair and compelling, and we maintain flexibility at the upper end of the range for exceptional candidates who can meaningfully accelerate our product and engineering roadmap.