Autonomous software development
Blitzy
Blitzy builds an autonomous software development platform for enterprise engineering teams creating applications and maintaining existing software. It handles work ranging from feature development and bug fixes to security improvements and documentation. The platform combines code generation with testing and validation, while human engineers define requirements and guide long-term software strategy.
Its defining technical choice is to spend more time and computation understanding a codebase and checking changes. Blitzy ingests source code into a relational index that preserves connections such as function calls, inheritance and dependencies between modules. Agents retrieve context using both semantic similarity and those relationships, helping connect a local edit to the surrounding system. Different models handle tasks such as generation and validation. The workflow creates tests, compiles changes and reruns existing tests, triggering further investigation when previously passing tests fail.
Brian Elliott and Sid Pardeshi began developing Blitzy while studying at Harvard, building it through the Harvard Innovation Labs and making it Pardeshi’s MS/MBA capstone project. Pardeshi brought six years of software engineering and architecture experience at NVIDIA. That background connects to the company’s focus on automating repetitive engineering work across the development process. Blitzy also places engineers with client companies to help integrate the platform into their work.
1 talk
Newest firstLoading more talks and speakers…
1 speaker at AIE
Affiliations reflect their AIE event listings and recorded appearances, not necessarily current employment.
