Shubham Sharma

As a 𝗦𝗼𝗳𝘁𝘄𝗮𝗿𝗲 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿, I advanced an 𝗲𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝗚𝗲𝗻𝗔𝗜 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺 by building 𝗮𝗴𝗲𝗻𝘁𝗶𝗰 𝗮𝗻𝗱 𝗺𝘂𝗹𝘁𝗶-𝗮𝗴𝗲𝗻𝘁 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 for research, analytics, and autonomous task execution. I owned 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄 𝗮𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲, 𝗮𝗴𝗲𝗻𝘁 𝗼𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻, 𝗺𝗼𝗱𝗲𝗹 𝗶𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻, 𝗟𝗟𝗠 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻, 𝗯𝗮𝗰𝗸𝗲𝗻𝗱 𝗔𝗣𝗜𝘀, 𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲 𝗼𝗽𝘁𝗶𝗺𝗶𝘇𝗮𝘁𝗶𝗼𝗻, 𝗱𝗲𝗽𝗹𝗼𝘆𝗺𝗲𝗻𝘁, 𝗮𝗻𝗱 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗺𝗼𝗻𝗶𝘁𝗼𝗿𝗶𝗻𝗴.

𝗥𝗲𝘃𝘀𝗔𝗜 – 𝗘𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝗚𝗲𝗻𝗔𝗜 𝗣𝗹𝗮𝘁𝗳𝗼𝗿𝗺
• Solved manual and fragmented AI workflows by extending the platform with agentic pipelines following a 𝗿𝗲𝘁𝗿𝗶𝗲𝘃𝗮𝗹 → 𝗿𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴 → 𝗮𝗰𝘁𝗶𝗼𝗻 pattern, improving task efficiency by ≈𝟰𝟬%.
• Built 𝗺𝘂𝗹𝘁𝗶-𝗮𝗴𝗲𝗻𝘁 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 that coordinated specialized agents for research, analytics, and autonomous execution, accelerating insight generation and complex task completion.
• Implemented 𝗺𝘂𝗹𝘁𝗶-𝗺𝗼𝗱𝗲𝗹 𝗼𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 across 𝗚𝗣𝗧-𝟰, 𝗖𝗹𝗮𝘂𝗱𝗲, 𝗮𝗻𝗱 𝗔𝗪𝗦 𝗕𝗲𝗱𝗿𝗼𝗰𝗸, optimizing model selection and reducing inference latency by ≈𝟯𝟬%.
• Developed 𝘀𝗰𝗮𝗹𝗮𝗯𝗹𝗲 𝗯𝗮𝗰𝗸𝗲𝗻𝗱 𝗶𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻𝘀, 𝗿𝗲𝘂𝘀𝗮𝗯𝗹𝗲 𝗮𝗴𝗲𝗻𝘁 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀, 𝘁𝗼𝗼𝗹-𝗰𝗮𝗹𝗹𝗶𝗻𝗴 𝗰𝗮𝗽𝗮𝗯𝗶𝗹𝗶𝘁𝗶𝗲𝘀, 𝗮𝗻𝗱 𝗺𝘂𝗹𝘁𝗶𝗺𝗼𝗱𝗮𝗹 𝗽𝗶𝗽𝗲𝗹𝗶𝗻𝗲𝘀 to support enterprise-wide adoption.
• Ensured 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗿𝗲𝗮𝗱𝗶𝗻𝗲𝘀𝘀 through evaluation, fallback handling, performance tuning, security controls, observability, and continuous optimization.

𝗔𝗰𝗰𝗼𝗺𝗽𝗹𝗶𝘀𝗵𝗺𝗲𝗻𝘁𝘀

• 𝗜𝗻𝗱𝗶𝗮 𝗔𝗜 𝗜𝗺𝗽𝗮𝗰𝘁 𝗦𝘂𝗺𝗺𝗶𝘁 & 𝗘𝘅𝗽𝗼 𝟮𝟬𝟮𝟲, 𝗕𝗵𝗮𝗿𝗮𝘁 𝗠𝗮𝗻𝗱𝗮𝗽𝗮𝗺: Participated in industry discussions and showcases focused on responsible AI, enterprise adoption, and real-world AI impact.
• 𝗡𝗔𝗦𝗦𝗖𝗢𝗠 𝗙𝘂𝘁𝘂𝗿𝗲 𝗙𝗼𝗿𝗴𝗲 & 𝗧𝗲𝗰𝗵 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿 𝗖𝗼𝗻𝗳𝗹𝘂𝗲𝗻𝗰𝗲 𝟮𝟬𝟮𝟱: Engaged with technology leaders and developers on deep-tech innovation, AI engineering, and scalable product development.

𝗧𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝗶𝗲𝘀: 𝗣𝘆𝘁𝗵𝗼𝗻, 𝗙𝗮𝘀𝘁𝗔𝗣𝗜, 𝗟𝗮𝗻𝗴𝗖𝗵𝗮𝗶𝗻, 𝗟𝗮𝗻𝗴𝗚𝗿𝗮𝗽𝗵, 𝗚𝗣𝗧-𝟰, 𝗖𝗹𝗮𝘂𝗱𝗲, 𝗔𝗺𝗮𝘇𝗼𝗻 𝗕𝗲𝗱𝗿𝗼𝗰𝗸, 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗔𝗜, 𝗠𝘂𝗹𝘁𝗶-𝗔𝗴𝗲𝗻𝘁 𝗦𝘆𝘀𝘁𝗲𝗺𝘀, 𝗥𝗘𝗦𝗧 𝗔𝗣𝗜𝘀, 𝗔𝗪𝗦, 𝗮𝗻𝗱 𝗠𝘂𝗹𝘁𝗶𝗺𝗼𝗱𝗮𝗹 𝗔𝗜

As an 𝗔𝘀𝘀𝗼𝗰𝗶𝗮𝘁𝗲 𝗦𝗼𝗳𝘁𝘄𝗮𝗿𝗲 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿, I delivered 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻-𝗿𝗲𝗮𝗱𝘆 𝗚𝗲𝗻𝗔𝗜, 𝗥𝗔𝗚, 𝗺𝘂𝗹𝘁𝗶-𝗮𝗴𝗲𝗻𝘁, 𝗮𝗻𝗱 𝗰𝗹𝗼𝘂𝗱 𝘀𝗼𝗹𝘂𝘁𝗶𝗼𝗻𝘀, owning system design, Python backends, retrieval and agent workflows, LLM integration and evaluation, REST APIs, optimization, deployment, and monitoring.

𝗥𝗲𝘃𝘀𝗔𝗜 – 𝗘𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝗚𝗲𝗻𝗔𝗜 𝗣𝗹𝗮𝘁𝗳𝗼𝗿𝗺

• Solved fragmented AI workflows by building a 𝘂𝗻𝗶𝗳𝗶𝗲𝗱 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺 combining 𝗺𝘂𝗹𝘁𝗶-𝗟𝗟𝗠 𝗰𝗵𝗮𝘁, 𝗥𝗔𝗚, 𝗲𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝘀𝗲𝗮𝗿𝗰𝗵, 𝗺𝘂𝗹𝘁𝗶𝗺𝗼𝗱𝗮𝗹 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲, 𝗮𝗻𝗱 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗲𝗱 𝗮𝗻𝗮𝗹𝘆𝘁𝗶𝗰𝘀.
• Developed 𝗿𝗲𝗮𝗹-𝘁𝗶𝗺𝗲 𝗺𝘂𝗹𝘁𝗶𝗺𝗼𝗱𝗮𝗹 𝗿𝗲𝘁𝗿𝗶𝗲𝘃𝗮𝗹 𝗽𝗶𝗽𝗲𝗹𝗶𝗻𝗲𝘀, improving accuracy by ≈𝟰𝟬% and reducing manual effort by ≈𝟲𝟬%.
• Designed 𝘀𝗰𝗮𝗹𝗮𝗯𝗹𝗲 𝗯𝗮𝗰𝗸𝗲𝗻𝗱 𝘀𝗲𝗿𝘃𝗶𝗰𝗲𝘀 𝗮𝗻𝗱 𝗮𝗻𝗮𝗹𝘆𝘁𝗶𝗰𝘀 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 that reduced analysis time by ≈𝟳𝟬%.

𝗦𝗺𝗮𝗿𝘁𝗔𝘀𝘀𝗶𝘀𝘁 – 𝗘𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝗔𝗜 𝗦𝘂𝗽𝗽𝗼𝗿𝘁 𝗖𝗼𝗽𝗶𝗹𝗼𝘁

• Solved slow and inconsistent knowledge discovery by building a 𝗥𝗔𝗚 𝗰𝗼𝗽𝗶𝗹𝗼𝘁 over 𝟭𝟬,𝟬𝟬𝟬+ 𝗲𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝗱𝗼𝗰𝘂𝗺𝗲𝗻𝘁𝘀.
• Owned 𝗶𝗻𝗴𝗲𝘀𝘁𝗶𝗼𝗻, 𝗰𝗵𝘂𝗻𝗸𝗶𝗻𝗴, 𝗲𝗺𝗯𝗲𝗱𝗱𝗶𝗻𝗴𝘀, 𝘃𝗲𝗰𝘁𝗼𝗿 𝗶𝗻𝗱𝗲𝘅𝗶𝗻𝗴, 𝗿𝗲𝘁𝗿𝗶𝗲𝘃𝗮𝗹, 𝗽𝗿𝗼𝗺𝗽𝘁 𝗱𝗲𝘀𝗶𝗴𝗻, 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻, 𝗔𝗣𝗜𝘀, 𝗱𝗲𝗽𝗹𝗼𝘆𝗺𝗲𝗻𝘁, 𝗮𝗻𝗱 𝗺𝗼𝗻𝗶𝘁𝗼𝗿𝗶𝗻𝗴.
• Achieved 𝟵𝟱%+ 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗲𝗱 𝗮𝗰𝗰𝘂𝗿𝗮𝗰𝘆, reduced resolution time by ≈𝟰𝟱–𝟲𝟬%, and improved productivity by 𝟮×.

𝗜𝗻𝘀𝗶𝗴𝗵𝘁𝗙𝗹𝗼𝘄 – 𝗖𝘂𝘀𝘁𝗼𝗺𝗲𝗿 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 𝗘𝗻𝗴𝗶𝗻𝗲

• Built a 𝗺𝘂𝗹𝘁𝗶-𝗮𝗴𝗲𝗻𝘁 𝗽𝗶𝗽𝗲𝗹𝗶𝗻𝗲 processing 𝟱𝟬,𝟬𝟬𝟬+ 𝗺𝗼𝗻𝘁𝗵𝗹𝘆 𝗶𝗻𝘁𝗲𝗿𝗮𝗰𝘁𝗶𝗼𝗻𝘀, achieving 𝟵𝟬%+ 𝗰𝗹𝗮𝘀𝘀𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗮𝗰𝗰𝘂𝗿𝗮𝗰𝘆 and reducing manual effort by ≈𝟳𝟬%.
• Improved routing and response time by ≈𝟰𝟬%; multimodal analysis increased insight coverage by ≈𝟯𝟬%.
• Owned 𝗮𝗴𝗲𝗻𝘁 𝗼𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻, 𝗶𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻𝘀, 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻, 𝘀𝗰𝗮𝗹𝗮𝗯𝗶𝗹𝗶𝘁𝘆, 𝗼𝗽𝘁𝗶𝗺𝗶𝘇𝗮𝘁𝗶𝗼𝗻, 𝗮𝗻𝗱 𝗺𝗼𝗻𝗶𝘁𝗼𝗿𝗶𝗻𝗴.

𝗔𝗰𝗰𝗼𝗺𝗽𝗹𝗶𝘀𝗵𝗺𝗲𝗻𝘁𝘀

• 𝗔𝗜 𝗛𝗮𝗰𝗸𝗮𝘁𝗵𝗼𝗻 𝟯.𝟬, 𝗜𝗜𝗧 𝗠𝗮𝗻𝗱𝗶: Led and evaluated teams developing GenAI solutions, including CSR-focused accessibility projects.
• 𝟴𝘁𝗵 𝗦𝘂𝗺𝗺𝗲𝗿 𝗦𝗰𝗵𝗼𝗼𝗹 𝗼𝗻 𝗔𝗜, 𝗜𝗜𝗜𝗧 𝗛𝘆𝗱𝗲𝗿𝗮𝗯𝗮𝗱: Completed expert-led sessions and hands-on projects covering advanced AI concepts.
• 𝗡𝗔𝗦𝗦𝗖𝗢𝗠 𝗔𝗜 𝗪𝗼𝗿𝗸𝘀𝗵𝗼𝗽: Explored applied AI, emerging technologies, and enterprise use cases.

𝗧𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝗶𝗲𝘀: 𝗣𝘆𝘁𝗵𝗼𝗻, 𝗙𝗮𝘀𝘁𝗔𝗣𝗜, 𝗗𝗷𝗮𝗻𝗴𝗼, 𝗟𝗮𝗻𝗴𝗖𝗵𝗮𝗶𝗻, 𝗟𝗮𝗻𝗴𝗚𝗿𝗮𝗽𝗵, 𝗢𝗽𝗲𝗻𝗔𝗜, 𝗖𝗹𝗮𝘂𝗱𝗲, 𝗚𝗲𝗺𝗶𝗻𝗶, 𝗔𝗺𝗮𝘇𝗼𝗻 𝗕𝗲𝗱𝗿𝗼𝗰𝗸, 𝗔𝘇𝘂𝗿𝗲 𝗢𝗽𝗲𝗻𝗔𝗜, 𝗩𝗲𝗰𝘁𝗼𝗿 𝗗𝗮𝘁𝗮𝗯𝗮𝘀𝗲𝘀, 𝗔𝗪𝗦, 𝗔𝘇𝘂𝗿𝗲, 𝗮𝗻𝗱 𝗚𝗖𝗣

• Developed a 3D web application using Flask and Three.js for electric bike battery pack design.
• Implemented packing optimization algorithms, improving space utilization by 25–35%.
• Built interactive 3D visualization and automated configuration, reducing design iteration time by ~40%.