# D.G. Suhaas Kiran > Founding AI Engineer, Previously at Graphite, Google, Infoblox Location: New York, New York, United States Profile: https://flows.cv/dg Currently pursuing an MS in Computer Science at the University of Massachusetts Amherst, while contributing as a Founding AI Engineer at a pre-seed stealth startup. Focused on personalized LLM applications, their work includes fine-tuning foundation models, building retrieval-augmented generation (RAG) pipelines, and deploying scalable LLM APIs. Academic projects at the DREAM and BioNLP Labs have further honed expertise in graph neural networks and vision-language models, emphasizing innovation and scalability. At the stealth startup, they have collaborated with the founding team to architect AI solutions and improve user-facing features. These experiences reflect a commitment to advancing AI research and practical applications. Passionate about driving efficient and scalable AI systems, they aim to integrate their technical expertise and collaborative approach to contribute to transformative projects in dynamic environments. ## Work Experience ### AI Software Engineer @ Olis Jan 2025 – Present | United States Part of a pre-seed startup focused on personalized LLM applications. Led core AI initiatives including fine-tuning foundation models, building RAG pipelines, and deploying scalable LLM APIs. Collaborated closely with founding team to shape the architecture and build the prototype. ### Graduate Student Researcher, DREAM Lab @ University of Massachusetts Amherst Jan 2024 – Present | Amherst, Massachusetts, United States - Conducted a literature survey on the existing time-series graph-based ML frameworks - TGN, TGL, NeutronStream - Developing a memory-state and dependency-graph based graph update mechanism to enable efficient and scalable training of graph neural networks on GPUs. - Implemented and benchmarked NeutronStream (Chaoyi Chen et al 2023) and ETC (Gao et al 2024) algorithms in DGL and PyG frameworks comparing link prediction accuracies and execution time with larger batch sizes. ### Graduate Student Researcher, BioNLP Lab @ University of Massachusetts Amherst Jan 2024 – Present | Amherst, Massachusetts, United States - Built baseline models for VWSD using InternVL2 VLM, and context augmentation using LLMs and lexical databases. - Experimented fine-grained image ranking with semantic similarity, Tree-of-Thought and knowledge graphs. - Developed a novel re-ranking method for fine-grained image-text matching using in-context learning, attribute decomposition and multi-step prompting. - Achieved an 8.7 percent points increase in top-1 accuracy for zero-shot matching and 0.8 points increase in MRR over fine-tuned SOTA ### Student Research Engineer @ University of Massachusetts Amherst Jan 2024 – Jan 2024 | Amherst, Massachusetts, United States - Built a keypoint detection model using YOLOv8 for automatic keypoint annotation of drone whale images - Implemented and tested heatmap-based keypoint detection models using UNet and ResNet. - Used active learning to fine-tune the model detections and reduce the initial annotation time by 80% ### Student Researcher Extern @ Graphite Jan 2025 – Jan 2025 | United States - Developed a multi-agent based long-form article generator using LangChain and RAG-based expert simulations for content generation. - Integrated FAISS for retrieval and implemented tool-calling pipelines to allow agents to access external documents and APIs. - Designed planner, writer and editor agents with feedback loops using tool calling and retrieval orchestration. - Developed a custom LLM-as-a-judge evaluation framework, achieving a 70% win rate over GPT-4o and a 5.04-point QA metric gain. ### Cloud Solutions Engineer -II (Data, AI/ML) @ Google Jan 2022 – Jan 2023 | Bengaluru, Karnataka, India - Troubleshot and optimized customer ETL pipelines and workflows, guiding integration of services like Dataflow, Composer, and Vertex AI through hands-on debugging and PoC development. - Provided proactive support for AI adoption by translating technical insights into customer-facing solutions. - Conducted knowledge sessions and workshops for Dataflow, Composer and Machine Learning. - Worked with product & sales teams for root cause analysis, to enhance product reliability, and build internal docs & productivity tools. - Built a machine-learning-based case sentiment analysis tool with 76% accuracy to proactively flag low-sentiment cases, reducing SLA misses by 70% and improving resolution time by 27%. - Led strategic customer engagement with Flipkart, ensuring high availability and proactive support during peak-scale events. ### Cloud Solutions Engineer (Data, AI/ML) @ Google Jan 2020 – Jan 2022 | Bengaluru, Karnataka, India ### Software Engineer Intern @ Infoblox Jan 2020 – Jan 2020 | Bengaluru Area, India Worked on identifying and removing the functionalities not related to the DNS for NIOS (Network Identity Operating System) and running tests to assess its effects on the performance of the system. ## Education ### Master of Science - MS in Computer Science University of Massachusetts Amherst Jan 2023 – Jan 2025 ### B.Tech in Electrical, Electronics and Communications Engineering PES University Jan 2016 – Jan 2020 ### Modern Public School, Bhiwadi Jan 2002 – Jan 2016 ## Contact & Social - LinkedIn: https://linkedin.com/in/d-g-suhaas-kiran-a304a1148 --- Source: https://flows.cv/dg JSON Resume: https://flows.cv/dg/resume.json Last updated: 2026-04-01