Data Scientist

Poiro (poiro.com) builds marketing AI systems and agents to bring brands closer to their customers.Poiro’s AI systems can be trained on both structured and unstructured marketing data for a brand - from social media content and e-commerce marketplace data to 1st-party customer data - to build a comprehensive knowledge representation of a brand and its category. On top of that, Poiro’s AI agents seamlessly perform data analytics & data science workflows to generate actionable insights and guide marketing execution.Poiro’s flagship product - Poiroscope - is one of the most advanced content creation suites for brands. Poiroscope enables brands to convert social research and insights into creative assets, without any prompting. Armed with proprietary models and computer vision workflows, Poiroscope is the only AI that generates consistently on-brand content with 100% accurate product capture.Leading brands are using Poiroscope to identify content whitespaces and generate highly engaging creatives across e-commerce, organic social and ads, at scale.Poiro is a subsidiary of Evam Labs (www.evamlabs.ai) - a Singapore headquartered holding company with offices in Bangalore and San Jose, building the next generation of high impact AI powered Enterprise Solutions. From Asia, for the World. Evam Labs was founded by ex-founders, academics and investors with over a decade of experience in building data & AI products and scaling companies from 0 to IPO. The founders are IIT/IIM/CMU alumni and have cumulatively raised over $500M+, invested in 30+ startups and hold 20+ patents. The rest of the Evam team comprises alumni of IITs, IIMs, CMU, IISc, NUS with rich experience across multiple industries.Role Overview -We are looking for a Data Scientist / Research Associate (Generative Marketing AI) to join us and make critical contributions to our product and solution development efforts. You’ll turn messy social signals into shippable content, competitive advantage, and growth. Zero fluff: prototype fast, ship often, measure impact. This is a full-time onsite role from our office located in HSR Layout, Bengaluru.Key Responsibilities -- Own the loop end-to-end: ideate → prototype → test → ship models/agents that power content generation, competitive analysis, and trend recommendations. - Build real data pipelines: ingest at scale from Instagram/TikTok/YouTube/X, web crawls, e-comm reviews, Google Trends; normalize with brand/product taxonomies; keep it reliable and cheap. - Model the marketing stack: - LLM + RAG for brief understanding, prompt planning, and insight summarization (brand docs, past campaigns, product specs). - VLMs/diffusion for creative scoring, brand-true checks (label integrity, color, composition), and content selection. - Recommenders for “creator × concept × timing” and trend fit. - Agentic workflows: stitch tools into agents that do the grunt work—brief-to-prompt, competitive watch, trend scout, concept A/B, asset QA, posting prep. - Experimental rigor: design offline metrics (recall@k, CLIP/aesthetic scores, accuracy on brand rules) and online tests (CTR, watch time, save/share rate, time-to-first-draft). - Productionize: serve models behind clean APIs, containerize, add tracing/monitoring, and set guardrails for cost/latency/reliability. - Evaluate open source: pick winners (models, vector DBs, schedulers) based on evidence; propose migrations or patches when they don’t meet a bar.About You -- Background: B.Tech in Computer Science (or equivalent). 1–2+ years hands-on in data/ML where things actually shipped. - Python first: strong with Pandas/Polars, PySpark; scikit-learn for quick baselines; you write clean, testable code. - Deep learning: PyTorch fluent (TensorFlow a plus). Comfortable fine-tuning, LoRA, and evaluation. - LLM/RAG: practical prompt engineering, tool-use/function-calling; you’ve built retrieval over docs with pgvector/Qdrant/Weaviate/Milvus and can defend your chunking/index choices. - Vision & gen-AI: experience with open-source VLMs (e.g., LLaVA, Qwen-VL, InternVL) and diffusion (SD/SDXL, ControlNet). Bonus: aesthetic/brand-safety classifiers, CLIP-based scoring. - Data at scale: Spark/Ray, Kafka/streams; solid SQL (Postgres/BigQuery). You know when to batch vs. stream. - Serving/infra: FastAPI, Docker, CI/CD; ideally Triton/TorchServe. Cloud literacy - Azure/AWS/GCP - ML ops: MLflow or Weights & Biases; reproducible experiments; sensible eval dashboards. - Marketing domain curiosity: creator economy, social platform quirks, what makes content “hit.” You like pairing models with real outcomes. - Mindset: resourceful, opinionated, allergic to hand-waving. You ship weekly, measure, and cut what doesn’t move the needle.Why Join Us?- Competitive salary and benefits - Be in an early stage company - grow with the company and shape our trajectory, in turn creating inorganic growth opportunities for yourself - Be in the right field at the right time - opportunity to grow and advance your career in the burgeoning field of AI - Be part of a fast-paced, ambitious research lab that is pushing the boundaries of AI, beginning with the marketing industry