Senior AI/ML Engineer

Experience : 5 years to 8 years

Locations: Delhi

5 Days working

Roles and Responsibilities

Data Foundation and Instrumentation

- Design and own the clickstream and impression-logging pipeline, including the

event taxonomy that every downstream model
- Normalise the catalogue and match products across gift cards, merchandise, and

services, deduplicating variants and building the taxonomy that co-occurrence statistics rely on Search and Retrieval
- Own search quality end to end: zero-result rate, query understanding, synonym and

typo handling, and relevance ranking
- Build semantic search and eligibility Q&A; that answers natural-language questions

such as “can I use my LSA for a gym membership,” grounded in the catalogue and plan-rule corpus, with citations, and abstains when the answer is not clear

Recommendations and Personalization

- Build co-purchase and cohort-based collections from order history, such as

“popular at your company,” “goes well with,” and occasion or points-tier curation
- Build session-based recommendations and in-session intent modelling, such as

“similar to what you’re viewing”
- Build personalized collections from browse and redemption behaviour
- Build a learned ranker over user, item, cohort, and context features, with eligibility,

balance, and availability enforced as hard constraints, not soft signals
- Build utilization intelligence that identifies who is at risk of forfeiting funds, what

they would plausibly spend on, and when to reach them Responsible and Privacy-Aware ML




- Enforce a hard architectural boundary around health-adjacent interaction data:

segregate it at the schema level and avoid modelling taste across it
- Treat confidently wrong eligibility answers as the worst failure mode, and design

retrieval and generation to say when they do not know instead of guessing

Serving and Reliability

- Deploy, serve, and operate models in production, owning latency, availability, and

monitoring for live systems

Experimentation and Business Impact

- Build offline evaluation that reflects online outcomes instead of eyeballed results
- Run A/B tests on utilization rate, order frequency, and search success, the metrics

clients renew on, not click-through rate alone
- Own and report the business outcomes of your work, connecting model and system

changes to measurable results Must-Haves
- 5+ years in production ML or ML-adjacent engineering
- Depth in at least one of search and ranking, or recommendations and

personalization, plus real interest in the other
- Shipped a production system from zero that you can walk us through in detail,

including what did not work
- Strong on embeddings and retrieval: vector search (FAISS, HNSW, pgvector, or a

managed equivalent),



hybrid lexical-plus-semantic retrieval, and reranking
- Comfortable with learned ranking, including GBDT rankers and neural or sequential

approaches where they earn their complexity
- Expert Python, robust SQL, and willing to own pipelines end to end (Spark or

equivalent, Airflow or Dagster or equivalent)
- Experience deploying and operating ML systems in production, including serving,

monitoring, and reliability of live models
- Rigorous about evaluation: you build offline eval sets instead of eyeballing outputs,

and you can explain how you would know your work created value
- Evidence of self-direction: you have set your own roadmap, negotiated scope with

stakeholders, and shipped without a manager breaking the work into tickets Nice-to-Haves
- RAG or grounded-generation systems in production, with a real strategy for

constraining hallucination
- Constrained ranking, where eligibility, budget, inventory, or regulatory filters sit over

model output
- Privacy-preserving or privacy-constrained ML, where data segregation or regulatory

boundaries shaped the design
- Entity resolution, product matching, or taxonomy work at scale
- Experience with Azure, ideally Azure Machine Learning
- Marketplace, insurance, healthcare, fintech, or loyalty experience, anywhere the

correctness of a rules decision mattered more than engagement
- Clickstream instrumentation designed from scratch
- Having been the first ML hire somewhere

Pay: ₹4,500,000.00 - ₹5,000,000.00 per year

Work Location: In person

📌 Senior AI ML Engineer (Delhi)
🏢 jobmatchmaker consultancy
📍 Delhi