Senior Software Engineering

Tejas Dave

As an experienced technology leader with a knack for tackling complex problems through

creative thinking, high spirits, and unwavering integrity in my work, I bring a collaborative spirit and get-it-done attitude to build invaluable assets for team environments. My varied experience spanning over 14 years and skillset ranging from AI/ML to ETL pipelines and Distributed Systems, equips me with the technical prowess to identify opportunities for improvement. I systematically work through a simple set of achievable goals to build out proficient end products. EXPERIENCE

Cisco Systems, San Francisco — Oct 2020 - Present

Senior Software Engineering @ Outshif by Cisco

Project Motific.ai: Advanced Retrieval-Augmented Generation (RAG) from scratch. Spearheaded the development team to integrate RAG to Enterprise Gateway where architecture caters to enterprise-data management and scalability.

• Key Achievements:

• Implemented a multitenant RAG system supporting 100,000 documents across 100 clients

• Integrated large language models with multimodal support, internal knowledge bases and data sources

• Improved accuracy and contextual relevance of generated responses using Precision and Recall

• Key role in developing Data ETL pipeline:

• Acquired and processed data via open telemetry connectors

• Implemented data cleaning and chunking techniques with Lang chain.

• Converted data to embeddings using SBERT/Mistral models

• Stored processed data in vector databases for efficient retrieval

• Developed system for user prompt processing:

• Implemented similarity search to fetch relevant data chunks

• Identified closest context based on search results

• Passed contextual information to Large Language Model for response generation

Technology: Large Language Models (LLM), LangChain, Apache Kafka MSK, Apache Spark - Amazon (EMR), PGvector, Kubernetes, Prometheus.io, Graphana, Huggingface · Sagemaker, FastAPI, Python3.11, Golang Senior Engineer @Emerging Technology and Venture Incubation Project Foresight: Build a microservice based training and validation pipeline that checks concept model drift or data drift using scouter models in production environment for any time series data in real time.

• Key Achievements:

• Addresses challenge of not having true label to compare model’s drif

• Enabled real time model validation in terms of accuracy, precession, f1 score & recall in production.

Technology: Sklearn, Kaggle, TensorFlow · Open Telemetry, Kafka, GRPC, MilvusDB, AWS S3, Golang

Work Auth: US Citizen

Home: San Diego, CA

Cell: +1-949-***-****

Email: ****.*****@*****.***

LinkedIn: www.linkedin.com/in/tejas-dave

Git: https://github.com/davetejas

SKILLS

Protocols:

KAFKA, RabitMQ, NFS, CIFS, NDMP, ZMQ,

GRPC, Protobuf

Databases

MongoDB, DynamoDB, Redis, Postgres,

Milvus, Vector DB, Level DB,

Tools

GDB, Docker, Kubernetes, Event bridge,

Valgrind, Kind, TensorFlow

Languages/Packages:

Golang, C/C++, Python3,

Apache Spark, PyTorch,

Langchain, Scikit-Learn,

Tensorflow, numpy, pandas

Platforms:

BSD 7.0, Ubuntu 12.04/14.0.4, GCP, AWS,

kaggle, huggingface

Distributed Systems:

Onefs, Ceph,

Cilantro (blockchain)

Education

MS Computer Science 2007

University of Southern California, USA

BE Computer Science 2004

VTU, Bangalore, India

Lamden INC, San Francisco — Sept 2018 - oct 2020

Lead Software Engineer - Cilantro Block Chain

● Developed distributed storage based on peer-to-peer architecture for ledger it is a SAAS feature which is hardware agnostic, highly reliable and consistent.

● Network upgrade/rollback protocol in untrusted environment

● Design and Implementation of Autonomous Governance Model for Cilantro blockchain nodes in untrusted network.

● Implemented distributed proxy for REST API’s, zmQ, grpc

● Built a Linter for smart contracts to enforce contract validity

● Deployment packages over AWS, Docker, GCP, Linux, Terraform

● Python libs: abstract syntax tree, zmQ, pymongo, capnp, sanic Dell EMC, Santa Clara CA/Seattle, WA — Dec 2011-Sept 2018 Sr Software Engineer - In Market Engineering

● Distributed File System for scale out storage solution – Isilon Onefs

● Worked on Snapshots, distributed locking, smart pools, backup/recovery features

● Volume Copy data migration File system migration tool using ndmp.

● Data Analytics on disk usage and storage patterns with system markers. It bypassed filesystem and read directly of super blocks

● Package manager for build and role out of Various releases to code.

● Auditor to analyse policies/access issues at protocol, filesystem, cluster layers Cisco Systems, San Jose CA — Aug 2007-Oct 2011

Software Engineer - Network Operating System Technology Group

● Core Routing/Switching software developer for Cisco's widely deployed Internet Operating System (IOS).

● Developed features, enhancements and optimizations such as BGP Fast Rerouting, Extending Graceful Restart (RFC4724), Cease Notifications (RFC4486), High Availability

Broadcom, Irvine, California — 2006-2007

Software Engineer Internship - High-Speed Controller Business Unit

● Profiling web services-based management (WS-Management) in OS-present and embedded environments.

● Performed packet flow analysis of Out-Of-Band management traffic in WS- Management and web-based environment.

● Evaluated performance for various hardware accelerations in the system. HEWLETT-PACKARD Co, Vancouver, WA — Summer 2006

Software Engineer Internship - Indigo Digital Press (IPG R&D)

● Performed prototyping of management capabilities with Indigo Digital press, which is comparable to private network of 15-25 computers with printing capability of 120 pages per min

ACADEMIC PROJECTS

P2P File Sharing Network:

Built Tracker & Peers with the objective of

tracking all peers interested in file sharing.

Each peer/node that is interested becomes

seeder. Also Eliminated the effect of free

loaders from network

Image Search Processor:

As Audio/Video data is imported and rendered

we extracted various semantics from it such as

dominant color, motion vector, energy levels.

This metadata was stored in DB and acted as a

search data set.

Spatial Database Project

University Transport System was developed

using spatial GUI in java, which supported

parsing and presentation of

spatial queries and spatial database was used

to model the geometric data in Oracle.

Stock Quote Extract

Used Perl script to sparse html source code for

yahoo finance to dynamically extract quote

regularly according to customs customized

settings. Values were stored and retrieved

from the background comma separated file.

ACADEMIC RESEARCH PAPERS

Security implication for web services.

Fall 2006,

Computer Security

Effect of new technologies on clustered

computing: RDMA, MPI, InfiniBand.

Fall 2006,

Advance O.S, Comp Arch

Contact this candidate