UMD CS B.S./M.S. · GRADUATING MAY 2027 · AWS SDE INTERN, SUMMER 2026
I build things from scratch and measure whether they work. Outside of work I'm into mechanical keyboards, rock climbing, and keeping up with tech.
Open to Summer 2027 new-grad software engineering roles
Three of the things I've built, with one measured result each. The names link to the code.
This summer I'm an SDE intern on EC2 Networking at AWS, where I built an agent that migrated seven production canaries from Scala to Java, cutting each one from about a week of work to about a day. Full timeline
Resume on request · ivanwang8989@gmail.com
A few things I spend my free time on.
You can customize every part, so you can tailor a board to exactly what you or someone else needs — and picking the right part for the job is the fun of it.
Key tester
logidle · no keys yetClick a key, or type while hoveringTap a key
I climb for the community and the problem solving. A boulder can usually be done a lot of different ways, and which one costs you the least depends on your body — working that out is my favorite part.
Boulder route
I keep up with tech pretty closely — from new hardware to how governments are handling AI. I'm interested in where the industry is heading.
What I follow
lodestone is a vector search engine I built in Rust. This is its benchmark: higher is more accurate, further right is faster, and each dot is one setting I measured.
readingef=32 · recall 0.9759 · ~31.8K q/s · single coredrag across the plot · or press ← →tap a point · drag to pan
50,000 x 128-d · k=10 · single core · Xeon 8488C · bench_50k_128d.txt (opens in new tab)
FIG. 02 · parallax-kv
DISTRIBUTED SYSTEMS · GO
A three-node Raft key-value store built from scratch: PreVote, ReadIndex, persist-before-send ordering, snapshots, a durable WAL, a gRPC runtime.
A deterministic stress harness ran 2.9 million client operations through partitions, crashes, message loss, delays, and snapshot churn. Zero safety or linearizability violations.
The first bug the stress harness found was in my checker, not in the consensus code.
Inspect the repository (opens in new tab)docs/BUG_LEDGER.md (opens in new tab)
3 nodes · localhost · c=8 · 4 KiB fsync p50 ≈ 0.89 ms
† The first bug the harness caught was in the linearizability checker, not in consensus. Not built: production snapshot scheduling or chunked streaming.
FIG. 03 · accretion-db
STORAGE ENGINE · RUST
An embeddable LSM-tree storage engine: CRC-framed WAL, group commit, memtables, block SSTables, bloom filters, size-tiered compaction. The crate root forbids unsafe code, so the engine has no unsafe blocks by construction.
Fault-injection and property tests confirm that acknowledged writes survive simulated crashes. Results are benchmarked against sled, including the cases where sled wins.
Inspect the repository (opens in new tab)benchmarks/RESULTS.md (opens in new tab)
16-byte keys / 100-byte values · ext4 on NVMe · Xeon 8488C
† The ~29x group-commit figure is WAL-bound. Compaction runs synchronously under the write lock, which caps full-engine throughput.
FIG. 04 · lodestone
VECTOR SEARCH · RUST
A vector search engine built from scratch in Rust for embeddings and RAG retrieval. An HNSW proximity graph and an IVF-PQ compressed index sit over hand-written AVX-512 distance kernels with runtime feature dispatch.
On 50,000 128-dimensional vectors on a single core, HNSW reached 0.976 recall@10 at about 31,800 queries per second — about 33x the exact brute-force search at that recall, and 9–33x across the 90%+ recall band.
Inspect the repository (opens in new tab)benchmarks/raw/bench_50k_128d.txt (opens in new tab)
50,000 x 128-d · k=10 · single core · Xeon 8488C
† IVF-PQ trades accuracy for footprint: 0.975 recall@10 at 16x memory compression. Single machine, single core. Not built: distributed sharding.
B.S./M.S. in Computer Science — Departmental Honors (research track)
On the faculty-mentored research track; minors in Data Science and Philosophy
Coursework: Data Structures & Algorithms, Object-Oriented Programming, Distributed Systems, Operating Systems, Computer Networks, Database Systems, Software Engineering, Machine Learning, Natural Language Processing
Python · Java · Go · Rust · TypeScript/JavaScript · Scala · SQL · Bash
Raft · gRPC/Protobuf · LSM trees · PostgreSQL/SQLite
Flask/FastAPI · Node.js · JUnit 5 · Playwright
PyTorch · HuggingFace · scikit-learn · LangChain · JAX · Qiskit · DisCoPy · pytket
AWS (EC2/VPC, SDK, CDK, CloudWatch) · Docker · Linux · GitHub Actions · CI/CD · Jenkins
I'm looking for Summer 2027 new-grad software engineering roles. Email is the fastest way to reach me, and I can send a full resume on request.