AI systems that turn your marketplace data into automated decisions
AI Engineer — Lead Scoring, Segmentation & Automated Outreach
-
Your best leads are already in your data — most teams just can't see them yet
-
I build the AI that scores, segments, and follows up automatically
-
Production-ready, properly evaluated, fast to deliver
The problem I solve
Most businesses have already invested in AI — a lead scoring model, an LLM pipeline, an automated workflow. But the model gets ignored because nobody trusts it. The pipeline produces inconsistent output. The automation fires at the wrong time. Nobody measured whether any of it actually works.
I find where your AI or sales process is leaking revenue, build the scoring and automation layer that plugs it, and validate it's working before I hand it over. That means lead scoring models your team actually relies on, personalised outreach that triggers at the right moment, and segmentation that makes your data actionable — not just present.
If you haven't built anything yet, I help you figure out what's worth building first.
What I build
- Lead Scoring
ML models that score every inbound lead by conversion likelihood — based on intent signals, behaviour, and interaction history. Your team stops guessing and starts prioritising.
- Customer Segmentation
Clustering systems that group customers by behaviour, value, and intent — so your messaging and sales effort targets the right people instead of everyone equally.
- Automated Outreach
LLM-powered email and SMS pipelines that trigger personalised follow-ups based on lead score, segment, or sales stage — the right message at the right time, without manual effort.
- Call Intelligence
Sales calls are your richest intent signal. I extract who was interested, what was said, and what was promised — and feed it back into your lead scoring and follow-up loop automatically.
About me
I'm Rimsha — an AI engineer who spent several years building production ML systems inside a large European online marketplace.
I've been in the room where business problems get defined. I work across the full cycle: understand the problem, scope what's worth building, design the right system, build it, evaluate whether it's actually working, and iterate until it is. Not just delivery — the whole thing.
I've built lead scoring, customer segmentation, listing content generation, pricing classification, automated communications, and call intelligence — all in production, across multiple markets and languages. I deliver fast, I scope tightly, and I'm honest about what's feasible before a line of code gets written.
I now work with marketplaces, SaaS companies, real estate platforms, and B2B sales teams that want AI that actually moves their numbers.
How an engagement works
- Discovery call
We talk through your business, where the manual work is, and what data you have available. I'll tell you honestly what's feasible and what isn't.
- Scoped proposal
I send a fixed-scope proposal: what I'll build, how long it takes, and what the deliverable looks like. No open-ended retainers, no scope creep.
- Build & deliver
I build the system, test it against your real data, and hand over working code with documentation. You own everything.
Tech stack
- Core
Python, FastAPI, Pydantic, Streamlit, Hydra
- ML / NLP
PyTorch, HuggingFace Transformers, scikit-learn, SpeechBrain, DeepEval
- LLMs & Agents
Claude, OpenAI, LangChain, LangGraph, AWS Bedrock, RAG, Prompt Engineering
- Cloud & Infrastructure
AWS (Bedrock, Athena, S3, EC2 GPU, ECR, CloudFormation), Docker, GitHub Actions
- Observability & Evaluation
Datadog, Langfuse, DeepEval, Deepchecks
- Data & Visualisation
Pandas, SQL, Plotly, Matplotlib, Streamlit
Frequently asked questions
What kinds of marketplace businesses do you work with?
I work with any online marketplace or platform that has listings, leads, and pricing data — automotive, property, B2B, rental, e-commerce. The problems I solve (lead scoring, listing quality, pricing signals, call intelligence) are domain-agnostic. If your business runs on listings and sales, the systems I build apply.
I have data but no ML team. Can you still help?
Yes, that's the typical engagement. You don't need an internal ML team. I assess your data, build the system, and hand over something your team can run — or I can help set up lightweight monitoring so it stays accurate over time.
What does a typical project cost and how long does it take?
Most focused projects (one system, one use case) land in the $5,000–$25,000 range and take 3–8 weeks. Scope and price depend on data availability, integration complexity, and whether you need ongoing evaluation. I scope everything upfront.
Will I own the code?
Yes. Everything I build is yours — source code, models, documentation. No vendor lock-in.
What if my data isn't clean or structured?
Real-world data rarely is. Data cleaning and pipeline work is part of what I do, not a blocker. I'll tell you during discovery if the data situation makes a project infeasible.
- Ready to talk?
Tell me what you're working on. If I can help, I'll say how. If I can't, I'll say that too.