1. Role Overview
We are seeking a Senior AI / Machine Learning Engineer to design and build systems that analyze points of interest (POIs) by extracting actionable insights from unstructured text, photos, and video media, and mapping them to dynamic user preferences and constraint-aware recommendations. This is a hands-on, execution-driven role focused on multimodal feature extraction, location APIs, spatial graph reasoning, and personalized recommendation systems.
2. Key Responsibilities
· API & Data Pipeline Engineering: Design robust ETL/ELT pipelines to ingest, cache, normalize, and deduplicate point-of-interest data, metadata, customer reviews, and multimedia assets from third-party APIs (such as Google Places API and open geospatial sources).
· NLP & Text Mining: Build pipelines for aspect-based sentiment analysis (ABSA), named entity extraction, and latent attribute mining across high-volume, unstructured user reviews.
· Computer Vision & Video Analysis: Implement models to process and score user-submitted images and short video frames (aesthetic scoring, scene classification, crowd/atmosphere detection, and visual tag generation).
· Multimodal Embedding & Retrieval: Align text, visual attributes, and spatial coordinates into unified vector spaces using modern multimodal architectures (e.g., CLIP, VLMs) to enable hybrid semantic search.
· Personalization & Routing Engines: Build recommendation algorithms and constraint-satisfaction solvers to rank, match, and sequence multi-stop geographic routes based on user preferences and physical travel constraints.
· Production Deployment & Scale: Containerize models, optimize inference latency and throughput, manage vector databases, and deploy production-grade microservices.
3. Technical Requirements & Skills Matrix
A. Machine Learning, NLP & Multimodal AI
· Deep Learning Frameworks: PyTorch or TensorFlow.
· NLP & LLMs: Hugging Face Transformers, Sentence-Transformers, spaCy, vLLM/Ollama, aspect-based sentiment analysis, prompt orchestration.
· Computer Vision: OpenCV, PIL, CLIP/OpenCLIP, BLIP/LLaVA or equivalent Vision-Language Models; practical experience in feature extraction, image tagging, and frame sampling from video (FFmpeg).
· Vector Search & Embeddings: Hands-on experience with vector databases (e.g., Qdrant, Milvus, Pinecone, or pgvector) and hybrid search (BM25 + dense retrieval).
B. Geospatial & Combinatorial Optimization
· Geospatial Processing: Solid grasp of spatial indexing, distance metrics, and libraries such as GeoPandas, Shapely, OSMnx, or PostGIS.
· API Integration: Production experience consuming third-party location platforms (Google Places API / Mapbox / Overpass API), including pagination, rate-limit management, and caching strategies.
· Constraint Optimization: Familiarity with routing/scheduling algorithms, graph traversal (NetworkX), or constraint solvers (e.g., Google OR-Tools, genetic algorithms, or heuristic search).
C. Software Engineering & Infrastructure
· Core Language: Expert-level Python (asynchronous programming, OOP, type hinting).
· APIs & Serving: FastAPI, Flask, or gRPC; experience with task queues (Celery, Redis, or Ray).
· Data Storage: PostgreSQL, Redis, and object storage (S3 / GCS).
· DevOps & CI/CD: Docker, basic Kubernetes, CI/CD pipelines, and cloud platform experience (AWS or GCP).
4. Experience & Qualifications
· Experience: 3+ years of hands-on experience building and deploying machine learning or data-intensive backend systems in production.
· Education: Bachelor’s or Master’s degree in Computer Science, Data Science, Electrical Engineering, Applied Mathematics, or equivalent practical experience.
· Track Record: Prior demonstrable work involving at least two of the following: location-based platforms, multimodal recommendation systems, large-scale review text mining, or automated sequencing/routing engines.
Pay: ₹539,077.94 - ₹1,866,715.20 per year
Application Question(s):
- How many years of hands-on experience do you have in Machine Learning/AI development and production deployment?
- Do you have hands-on experience with Python and ML frameworks such as PyTorch or TensorFlow?
- Have you worked with location/geospatial APIs such as Google Places, Mapbox, or OpenStreetMap/Overpass API?
Work Location: In person