Somak S.

Agentic AI | Automation | Backend | Fullstack Engineer

𝗧𝗼𝗽 𝗥𝗮𝘁𝗲𝗱 𝗔𝗜 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿 & 𝗙𝘂𝗹𝗹-𝗦𝘁𝗮𝗰𝗸 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿 | 𝟴+ 𝗬𝗲𝗮𝗿𝘀 | 𝟭% 𝗼𝗳 𝗨𝗽𝘄𝗼𝗿𝗸 | 𝟭𝟬𝟬% 𝗝𝗼𝗯 𝗦𝘂𝗰𝗰𝗲𝘀𝘀 ✅ $30K+ Total Earnings ✅ 8+ Years as a Full-Stack Developer ✅ 80+ Projects Delivered ✅ Top Rated Plus ✅ 100% Job Success Score ✅ AWS Certified | GCP Certified | Python Certified ✅ 50 hrs/week available ✅ 4+ AI/ML production integrations 🔴 Ranked in the 𝗧𝗼𝗽 𝟭% on Upwork. 🔴 Ranked in the 𝗧𝗼𝗽 𝟰% on Stack Overflow. 🔴 B.Tech Computer Science & Engineering — IIIT Gwalior. 𝐀𝐈 𝐀𝐠𝐞𝐧𝐭𝐬 / 𝐕𝐨𝐢𝐜𝐞 𝐀𝐠𝐞𝐧𝐭𝐬: 𝐂𝐫𝐞𝐰𝐀𝐈 / 𝐀𝐮𝐭𝐨𝐆𝐞𝐧 / 𝐀𝐦𝐚𝐳𝐨𝐧 𝐏𝐨𝐥𝐥𝐲 / 𝐃𝐞𝐞𝐩𝐠𝐫𝐚𝐦 / 𝐑𝐚𝐬𝐚 𝐀𝐈 / 𝐂𝐥𝐚𝐮𝐝𝐞 / 𝐂𝐨𝐝𝐞𝐱 / 𝐂𝐨𝐰𝐨𝐫𝐤 / 𝐀𝐳𝐮𝐫𝐞 𝐀𝐈 𝐒𝐩𝐞𝐞𝐜𝐡 / 𝐋𝐋𝐌 𝐅𝐢𝐧𝐞𝐭𝐮𝐧𝐢𝐧𝐠: 𝐏𝐄𝐅𝐓 / 𝐑𝐋𝐇𝐅 / 𝐃𝐏𝐎 / 𝐒𝐅𝐓 𝐰𝐢𝐭𝐡 𝐔𝐧𝐬𝐥𝐨𝐭𝐡 / 𝐀𝐱𝐨𝐥𝐨𝐭𝐥 / 𝐇𝐮𝐠𝐠𝐢𝐧𝐠𝐅𝐚𝐜𝐞 𝐀𝐮𝐭𝐨𝐓𝐫𝐚𝐢𝐧 / 𝐒𝐚𝐠𝐞𝐌𝐚𝐤𝐞𝐫 𝐓𝐫𝐚𝐢𝐧𝐢𝐧𝐠 / 𝐎𝐩𝐞𝐧-𝐒𝐨𝐮𝐫𝐜𝐞 𝐋𝐋𝐌𝐬: 𝐋𝐋𝐀𝐌𝐀 𝟑 / 𝐌𝐢𝐬𝐭𝐫𝐚𝐥 𝟕𝐁 / 𝐌𝐢𝐱𝐭𝐫𝐚𝐥 𝟖𝐱𝟕𝐁 / 𝐅𝐚𝐥𝐜𝐨𝐧 / 𝐆𝐞𝐦𝐦𝐚 / 𝐅𝐚𝐬𝐭 𝐈𝐧𝐟𝐞𝐫𝐞𝐧𝐜𝐞: 𝐯𝐋𝐋𝐌 / 𝐓𝐆𝐈 / 𝐓𝐞𝐧𝐬𝐨𝐫𝐑𝐓-𝐋𝐋𝐌 / 𝐐𝐮𝐚𝐧𝐭𝐢𝐳𝐚𝐭𝐢𝐨𝐧: 𝐀𝐖𝐐 / 𝐆𝐏𝐓𝐐 / 𝐆𝐆𝐔𝐅 / 𝐆𝐆𝐌𝐋 / 𝐐𝐋𝐎𝐑𝐀 / 𝐑𝐀𝐆 𝐒𝐲𝐬𝐭𝐞𝐦𝐬: 𝐋𝐚𝐧𝐠𝐂𝐡𝐚𝐢𝐧 / 𝐋𝐥𝐚𝐦𝐚𝐈𝐧𝐝𝐞𝐱 / 𝐂𝐡𝐫𝐨𝐦𝐚 / 𝐅𝐀𝐈𝐒𝐒 / 𝐏𝐢𝐧𝐞𝐜𝐨𝐧𝐞 / 𝐐𝐝𝐫𝐚𝐧𝐭 / 𝐖𝐞𝐚𝐯𝐢𝐚𝐭𝐞 / 𝐌𝐢𝐥𝐯𝐮𝐬 / 𝐀𝐥𝐠𝐨𝐥𝐢𝐚 / 𝐕𝐞𝐫𝐭𝐞𝐱 𝐀𝐈 / Google Cloud Agent Platform / Agent Runtime / Agent Memory Bank Hello! I'm Somak Somanchi — a seasoned full-stack and AI engineer with 8+ years of hands-on experience building web applications, cloud-native systems, and production-grade AI integrations. I've spent the last several years deep in LLM territory — designing AI agents with LangChain and LlamaIndex, building RAG pipelines on top of vector stores, and shipping integrations with Claude, GPT-4o, Amazon Bedrock, and Ollama into real customer-facing products. 🌟 𝗔𝗱𝘃𝗮𝗻𝗰𝗲𝗱 𝗔𝗜 & 𝗟𝗟𝗠 𝗦𝘁𝗮𝗰𝗸: 🔹 AI Agents / Voice Agents: CrewAI, AutoGen, Amazon Polly, Deepgram, Rasa AI, Claude, Codex, Cowork 🔹 Open-Source LLMs: LLaMA 3, Mistral 7B, Mixtral 8×7B, Falcon, Gemma, Bloom, Orca Mini, Guanaco 🔹 Inference Optimization: vLLM, TGI, TensorRT-LLM, SKPilot 🔹 Prompt Engineering: Multi-turn, Few-shot, Zero-shot, RAG-based, Programmable Pipelines 🔹 Quantization: AWQ, GPTQ, GGUF, GGML, QLoRA, PTQ, DQ 🔹 RAG Systems & Vector DBs: LangChain, LlamaIndex, Chroma, FAISS, Pinecone, Qdrant, Weaviate, Milvus, Algolia 🔹 Search Infrastructure: Vector Search, Autocomplete Search, Search Index Building 🔹 LLM Deployment: AWS SageMaker, Vertex AI Platform, RunPod, GCP AI Platform, Vercel AI SDK 🔹 Data Pipelines: Synthetic dataset generation, LLM evaluation frameworks 🖥️ 𝗕𝗮𝗰𝗸𝗲𝗻𝗱 𝗦𝗸𝗶𝗹𝗹𝘀: 🔹 Node.js, Express.js, Nest.js, Python, Django, Flask, FastAPI, Java, Spring Boot MVC, Golang 🔹 REST API, GraphQL, WebSockets, Real-time Chat (Ably) 🔹 Relational & NoSQL: PostgreSQL, MySQL, MongoDB, Firebase, Firestore, Supabase, Amazon Aurora & RDS 🔹 Event streaming with Kafka; task queues with Celery 🔹 Xano (no-code backend) for rapid MVP delivery 🔹 LangChain, LangServe, LangSmith, HuggingFace Transformers for AI/LLM integrations 🔹 Low-code AI tools: Flowise AI, LangFlow, StackAI for rapid prototyping 🔹 Testing & automation: Pytest, Unittest, Selenium 🌐 𝗙𝗿𝗼𝗻𝘁𝗲𝗻𝗱 𝗦𝗸𝗶𝗹𝗹𝘀: 🔹 TypeScript, React.js, Next.js, Vue.js, Nuxt.js, React Native 🔹 Redux Toolkit for state management 🔹 Tailwind CSS, HTML5, CSS3, responsive UI/UX design 🔹 Progressive Web Apps (PWA) and Single Page Applications (SPA) ☁️ 𝗖𝗹𝗼𝘂𝗱 & 𝗗𝗲𝘃𝗢𝗽𝘀: 🔹 AWS: Lambda, Fargate, Step Functions, EC2, S3, SageMaker, RDS, Aurora 🔹 VPC Management, Subnet Grouping, NAT Gateway configuration 🔹 Google Cloud Platform (Certified), Azure 🔹 Docker, Kubernetes, Redis, Nginx for scalable infrastructure 🔹 CI/CD pipelines and containerized deployments 🔌 𝗜𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻𝘀 & 𝗔𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗼𝗻: 🔹 Social Media Automation pipelines (multi-platform publishing) 🔹 Mobile Proxy-based Scrapers for compliant data collection 🔹 Stripe API for payments, Outlook & Calendar integrations 🔹 Google Maps & Google Places API 🔹 Zapier (3+ yrs), JIRA (4+ yrs), Airtable, Productive, QuickBooks 🔹 ResourceSpace DAM, Glitch deployments 🛠️ 𝗧𝗼𝗼𝗹𝘀 & 𝗖𝗼𝗿𝗲 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲𝘀: 🔹 Python ML libraries: Scikit-learn, NumPy, Pandas, Matplotlib, Seaborn 🔹 OpenAI APIs, Whisper, GPT models, ChatGPT integration, AI chatbot deployment 🔹 Python (3+ yrs), Java (3+ yrs), C/C++ (5+ yrs), Golang (2+ yrs) 🔹 Git/GitHub, Linux environments (Ubuntu, CentOS) My approach to every project starts the same way — making sure the client and I share a clear picture of the goal, the constraints, and what "done" actually looks like. From there, I bring architectural rigor and a builder's instinct: clean code, sensible cloud topology, observability built in, and AI integrations.