Job Title

About OpptraOpptra, founded by Binny Bansal (Flipkart co-founder), is transforming global e-commerce by enabling consumer brands to scale across Asia’s 2.4 billion-consumer market and beyond. Our franchising portfolio—Exporio (fashion), Terraspan (home & kitchen), and a forthcoming electronics division—ispowered by our AI Growth. We’re deploying AI agents to revolutionize workflows like pricing, forecasting, cataloging, and marketing, and we need an AI Agent Builder to design, build, and scale these systems rapidly, delivering real-world impact for global consumer brands.Why Join Opptra?Impact: Build agents that power global consumer brands in transforminghow millions shop across Asia.AI-First Culture: Work with the latest tools—LangChain, Claude, Pinecone—in a team obsessed with pushing AI boundaries.Vision: Be part of Binny Bansal’s mission to redefine global commerce Growth: Join a company scaling from 45 to 125 in 2025, with opportunities to lead agent initiatives and shape Opptra’s AI stack. Flexibility: Thrive in a remote-first setup with async workflows tailored forAsia time zones.Role OverviewAs an AI Agent Builder at Opptra, you’ll create and deploy intelligent AI agents that power our e-commerce platforms, automating critical workflows such as dynamic pricing, demand forecasting, product cataloging, and personalized marketing. Using cutting-edge tools like LangChain, CrewAI, AutoGen, and models like GPT-4o, Claude 3, and Mixtral, you’ll build agents that deliver results in a week, test them live with brands, and iterate relentlessly. You’ll write precise prompts, leverage vector databases like Pinecone and Weaviate, anddevelop reusable templates to ensure scalability. Collaborating directly with brand teams—not just engineers—you’ll translate business needs into AI solutions, embedding agents into platforms like Shopify, Shopee, and Amazon to drive efficiency and growth across Asia’s markets.Key OutcomesAgent Development:Build autonomous AI agents using LangChain, CrewAI, and AutoGen to handle tasks like pricing optimization, inventory forecasting, and marketing campaign generation. Integrate advanced LLMs (GPT-4o, Claude 3, Gemini Ultra, Mixtral) via APIs like Fireworks.ai or Anthropic for natural language tasks, such asgenerating product descriptions or customer responses. Develop vision-based catalog agents with GPT-4 Vision, BLIP, or CLIP to automate image recognition, tagging, and categorization for millionsof SKUs. Write high-quality prompts that achieve specific outcomes (e.g., 90% accuracy in pricing recommendations), using techniques like chain-of- thought and few-shot learning.Vector Search and RAG:Implement retrieval-augmented generation (RAG) pipelines with vector databases (Pinecone, Weaviate, FAISS) to power personalized search, recommendations, and trend analysis. Optimize embeddings for multi-language support (e.g., Hindi, Arabic, Bahasa) using SentenceTransformers or multilingual BERT, ensuringrelevance across India, GCC, and SEA. Build knowledge bases for agents, integrating structured (SQL,Snowflake) and unstructured (Notion, Google Drive) data for real-time decision-making.Rapid Prototyping and Iteration: Deliver functional agents in one-week sprints, using Python (Pandas,NumPy) and Jupyter for prototyping and testing. Deploy agents to live environments (e.g., Shopify stores, Amazon listings), monitoring performance with tools like Weights & Biases andLangSmith. Iterate based on user feedback from brand teams, achieving metrics like 20% faster catalog updates or 15% higher conversion rates. Create evaluation frameworks (e.g., BLEU, ROUGE for text; precision/recall for vision) to benchmark agent performance and log results for traceability.Template and Framework Creation:Develop reusable agent templates for common e-commerce tasks (e.g., pricing, A/B testing), using AutoGen or CrewAI to modularize workflows. Build logging systems to track agent actions and errors, integrating withGrafana or Datadog for real-time insights. Document templates and prompts in a centralized repo (Notion,GitHub), enabling brand teams to customize agents without engineering support.Collaboration with Brands:Work directly with brand teams (Exporio, Terraspan) to understand needs, translating requirements into agent capabilities (e.g., automating GCC fashion campaigns). Partner with product managers and engineers to integrate agents into platforms like Shopify, Shopee, and Amazon, ensuring compatibilitywith APIs and webhooks. Train non-technical stakeholders on agent usage via workshops, using tools like Streamlit or Gradio for interactive demos.Scalability and Optimization:Optimize agents for low-latency inference (sub-500ms) using techniques like quantization (TensorRT, ONNX) and caching (Redis). Deploy agents at scale with MLOps tools (MLflow, Kubeflow) on AWS (Bedrock, SageMaker) or GCP (Vertex AI), handling 1M+ daily transactions.Ensure compliance with regional data laws (e.g., India’s DPDP Act, UAE’s PDPL) using encryption (AES-256) and secure APIs (OAuth 2.0).CompetenciesExperience:3–6 years working with large language models (LLMs) and generative AI, with a focus on building functional agents in production environments. Proven track record of shipping AI agents that solve real-worldproblems, such as automation in e-commerce, retail, or marketing.Deep experience with e-commerce workflows, including pricing, catalog management, or campaign optimization.AI Proficiency:Expertise in agent frameworks: LangChain, CrewAI, AutoGen for building multi-agent systems and workflows.Proficiency with LLMs: GPT-4o, Claude 3, Gemini Ultra, Mixtral, using APIs like Fireworks.ai, Anthropic, or Google Cloud. Experience with vision models: GPT-4 Vision, BLIP, or CLIP for catalog automation and image analysis.Strong skills in retrieval-augmented generation (RAG) and vector search, using Pinecone, Weaviate, or FAISS for embeddings and knowledge retrieval. Advanced prompt engineering, leveraging chain-of-thought, few-shot,and zero-shot techniques to achieve 85%+ task accuracy.Technical Skills:Expert in Python (Pandas, NumPy, FastAPI) for agent development, with familiarity in TypeScript or Go for API integration. Experience with vector databases (Pinecone, Weaviate, FAISS) andSQL/NoSQL (PostgreSQL, MongoDB) for data pipelines. Proficiency in MLOps tools (MLflow, Kubeflow, Weights & Biases) for model deployment, monitoring, and versioning. Knowledge of cloud platforms (AWS, GCP, Azure) and containerization (Docker, Kubernetes) for scalable agent hosting.Problem-Solving and Taste:Ability to design agents that are practical and impactful, balancing complexity with usability (e.g., intuitive outputs for brand teams). Strong debugging skills, using tools like LangSmith or Sentry to resolve agent failures in live environments. Keen sense of “what works,” prioritizing solutions that drive measurable outcomes (e.g., 30% faster workflows).Work Style:Comfortable in a remote-first, fast-paced environment, managing async collaboration across Asia time zones (IST, GST, SGT). Effective communicator, able to explain agent functionality to brandteams via Slack, Zoom, or Notion. Bias for action, thriving in one-week build-test-iterate cycles under tightdeadlines.Preferred QualificationsE-commerce Expertise:Experience shipping AI solutions on platforms like Shopify, Shopee, or Amazon, with familiarity in APIs (Shopify GraphQL, Amazon SP-API). Knowledge of Asian e-commerce trends, such as quick-commerce(Blinkit, Grab) or cross-border logistics (Aramex, DHL).Advanced AI Skills:Expertise in fine-tuning LLMs with tools like Unsloth, LoRA, or TRL for domain-specific tasks (e.g., fashion cataloging).Experience with multimodal agents, combining vision (YOLO, OpenCV)and text (Llama 3) for tasks like product quality checks. Familiarity with reinforcement learning (Ray RLlib, Stable-Baselines3) for optimizing agent decision-making.Innovation Track Record:Contributions to open-source AI projects (e.g., LangChain, Hugging Face) or published blogs on agent development. Experience scaling agents for high-traffic environments (>1M dailyinteractions), such as Black Friday sales or product launches. Knowledge of emerging AI protocols like A2A (Agent-to-Agent) or MCP for cross-agent orchestration.AI Tools You’ll UseAgent Frameworks: LangChain, CrewAI, AutoGen for multi-agent systems. LLMs: GPT-4o, Claude 3, Gemini Ultra, Mixtral via Fireworks.ai, Anthropic, Google Cloud. Vision Models: GPT-4 Vision, BLIP, CLIP for catalog and image tasks. Vector Databases: Pinecone, Weaviate, FAISS for RAG and search.Development: Jupyter, VS Code, PyCharm for coding; Postman for API testing.MLOps: MLflow, Kubeflow, Weights & Biases, LangSmith for deploymentand monitoring. Data: Snowflake, PostgreSQL, MongoDB for analytics and pipelines.Infrastructure: AWS (Bedrock, SageMaker), GCP (Vertex AI), Docker forhosting.Collaboration: Notion, Slack, Figma, Linear for specs and communication.