Applied ML Engineer

remotestar-team; Gurugram District; On-site; INR 2,500,000 - 4,000,000; Full time

Job summary

remotestar-team’s client in Gurgaon is seeking an Applied ML Engineer to own the recommender system powering in‑game commerce. You will leverage 3–5 years of ML engineering experience with a focus on collaborative filtering, embeddings, and production evaluation.

Strong Python (PyTorch/JAX) and experience moving models from offline to online serving are required. You will work closely with Data and Backend Engineers to tune embeddings, enable Faiss-based retrieval, and drive improvements in CTR,

Qualifications

• 3–5 years of ML engineering experience, with recommender systems specifically.
• Strong Python: PyTorch / JAX, scikit‑learn, NumPy.
• Hands‑on with collaborative filtering — sparse matrices, cold start, production evaluation.
• Embedding‑based retrieval experience (Faiss, ScaNN, or equivalent).

Responsibilities

• Own the collaborative filtering model (starting with Gorse, potentially moving to a custom stack).
• Build product embeddings (product2vec + Faiss / ANN) for the PlaySuper catalogue.
• Evolve cohort assignment from rules‑based to ML‑driven.
• Build the offline evaluation framework — precision@k, NDCG, conversion‑rate, diversity, coverage.
• Bridge offline models to online serving (model serving infrastructure, weekly refresh pipeline).
• Calibrate ranking weights against business outcomes (CTR, GMV, margin, repeat redemption).
• Partner closely with the Data Engineer (event pipeline + feature store) and Backend Engineer (ranking API, Redis serving layer).
• Translate sparse, noisy in‑game event data into reliable signal.
• Act as the internal owner of recommendation quality — always pushing on whether the model is actually lifting outcomes vs. just looking good on offline metrics.

Skills

Python

PyTorch

JAX

scikit-learn

NumPy

Collaborative filtering

Embedding-based retrieval

Faiss

Gorse

LightFM

Tools

Faiss

ScaNN

Gorse

Job description

Role: Applied ML Engineer

Location: Gurgaon (on‑site) 5 days WFO

Employment type: Full‑time

Company Overview:

Our client is revolutionising the gaming landscape as India’s first G-Commerce startup, bridging the gap between virtual achievements and real‑world value. We empower gamers by transforming their in‑game currencies and XP into tangible rewards, from exclusive brand discounts to physical goods, lowering cart values and making gaming more rewarding than ever.

Through strategic partnerships with game developers and top‑tier brands, we create seamless white‑labeled reward ecosystems, integrated directly into games and gaming platforms. Our mission? To redefine engagement by turning every play session into an opportunity for players to earn, redeem, and experience more.

Role Overview:

We are looking for an Applied ML Engineer with strong experience in recommender systems to build the brain of PlaySuper's in‑game commerce store — a recommendation engine that decides which products, coupons, and rewards to surface to which player, at which moment.

This role is ideal for someone who has worked on:

• Collaborative filtering and embedding‑based retrieval in production


• Recommendation systems for marketplaces, deals, or content feeds


• Cold‑start and sparse‑data problems


• Bridging offline model development to online serving

Rigorous evaluation and a bias for shipping are non‑negotiable.

What You Will Do:

• Own the collaborative filtering model (starting with Gorse, potentially moving to a custom stack)


• Build product embeddings (product2vec + Faiss / ANN) for the PlaySuper catalogue


• Evolve cohort assignment from rules‑based to ML‑driven


• Build the offline evaluation framework — precision@k, NDCG, conversion‑rate, diversity, coverage


• Bridge offline models to online serving (model serving infrastructure, weekly refresh pipeline)


• Calibrate ranking weights against business outcomes (CTR, GMV, margin, repeat redemption)


• Partner closely with the Data Engineer (event pipeline + feature store) and Backend Engineer (ranking API, Redis serving layer)


• Translate sparse, noisy in‑game event data into reliable signal


• Act as the internal owner of recommendation quality — always pushing on whether the model is actually lifting outcomes vs. just looking good on offline metrics

What We Are Looking For:

• 3–5 years of ML engineering experience, with recommender systems specifically


• Strong Python: PyTorch / JAX, scikit‑learn, NumPy


• Hands‑on with collaborative filtering — sparse matrices, cold start, production evaluation


• Embedding‑based retrieval experience (Faiss, ScaNN, or equivalent)


• Proper reco evaluation chops — beyond accuracy: diversity, coverage, business‑outcome metrics


• Comfort with sparse and noisy data


• Experience taking models from offline notebooks to online serving in production


• Clear communication and structured problem‑solving

Strong Plus (Nice to Have)

• Experience building voucher, coupon, or deal recommendation systems


• Gaming, mobile, or consumer‑engagement product experience


• Familiarity with Gorse or LightFM


• Experience with contextual bandits or online learning


• Feature store patterns


• Startup experience or ownership in fast‑moving environments