I

Ishika Kulkarni

About

Detail

California, United States

Timeline

work
Job
school
Education

Résumé

Jobs verified_user 0% verified
  • Juniper Square
    Artificial Intelligence Engineering Intern
    Juniper Square
    Jun 2025 - Current (1 year 5 months)
    • Built and deployed a production-grade multi-agent system on MCP servers, integrating AWS Bedrock services, external APIs, and a complex multi-turn RAG pipeline that scaled to real-world financial data queries and improved conversational accuracy. • Prototyped and integrated LangSmith for 100% tool coverage, implementing token usage tracking and cost estimation dashboards that enabled data-driven model selection and cut inference costs by 25% without com promising reliability. • Collaborated on production deployment and validation of the multi-agent framework, strengthening workflow orchestration and ensuring end-to-end robustness, reliability, and continuity in extended user conversations.
  • Uniphore
    Artificial Intelligence Researcher Apprenticeship
    Uniphore
    Apr 2025 - Current (1 year 7 months)
    • Built a multi-turn conversational system with dynamic tool calling and adaptive agent behaviors, generating a custom dataset from these interactions to enable more realistic evaluation and robust agent development. • Applied reinforcement learning (RL) to fine-tune LLMs and designed orchestration workflows, ensuring re sponses aligned with user needs while establishing reproducible benchmarks for future system improvements.
  • S
    Software Developer
    Signify Innovation Labs (Formerly Philips Lighting)
    Jul 2023 - Aug 2024 (1 year 2 months)
    • Authored Android SDK with 30+ REST APIs for IoT light/sensor control, adopted in 5+ production systems, reducing code redundancy by 35%. • Integrated Zigbee communication protocol across smart apps, reducing customer-reported connectivity issues.
Education verified_user 0% verified
  • University of California, Santa Cruz
    Master of Science in
    University of California, Santa Cruz
    Sep 2024 - Dec 2025 (1 year 4 months)
  • Vellore Institute of Technology
    Bachelor of Technology in
    Vellore Institute of Technology
    Jun 2019 - May 2023 (4 years)
Projects (professional or personal) verified_user 0% verified
  • L
    LLM Pushback Evaluation Framework Group Project
    – Established a novel LLM evaluation framework with user feedback data and model refinement strategies. – Evaluated LLM response quality across 10 distinct prompt categories, identifying the most effective strategies for boosting user trust and reducing instances of inaccurate or unhelpful outputs.
  • C
    ChartQA-VLM: Fine-tuning Qwen2-VL-7B for Chart Question Answering Personal
    – Deployed Transformer Reinforcement Learning model utilizing reward-based training to enhance factual ac curacy in visual chart Question Answering by 10%.