Developer Tools●●Solid
Otters – A Pandas-style DataFrame library written in pure Go
Pandas API in pure Go, but Polars and DuckDB already own this space.
Big BrainNiche GemShip It
fahadishere
313mo ago
Back to browse
GitHub Repository
Drop-in pandas replacement backed by PyTorch custom C++ ops — enables torch.jit.script compilation and pure C++ inference
4 starsPython
Xpandas – running Pandas-style computation directly in pure C++
by xqli·Mar 2, 2026·6 points·0 comments
AI Analysis
●●●BangerWizardryBig BrainZero to One
Compiles pandas logic to TorchScript, then runs C++-only inference—eliminates Python runtime bottleneck for quant workflows.
Strengths
- •Novel deployment pattern: write pandas in Python, freeze to .pt, ship to bare-metal C++ with zero rewrite—GIL and latency disappear entirely.
- •35 ops (groupby, rolling, ewm, shift, zscore, cumulative) cover 80% of quant signal engineering without reimplementing all of pandas.
- •Benchmarks show 2–163× speedup on rolling and element-wise ops; pure C++ inference removes Python interpreter overhead and GIL blocking.
Weaknesses
- •Subset of 35 ops limits adoption to specific domains (quantitative trading, online ML inference); general data science users still need pandas.
- •Immature ecosystem—no integrations, raw experimental quality, and depends on PyTorch's C++ bindings (adds deployment complexity).
Category
Target Audience
Quantitative traders, high-frequency trading engineers, edge/embedded ML teams, low-latency inference platforms
Similar To
Polars · Modin · Datafusion
Post Description
Hi HN,
I’ve been exploring whether pandas can be used as a computation description, rather than a runtime.
The idea is to write data logic in pandas / NumPy, then freeze that logic into a static compute graph and execute it in pure C++, without embedding Python.
This is not about reimplementing pandas or speeding up Python. It’s about situations where pandas-style logic is useful, but Python itself becomes a liability (latency, embedding, deployment).
The project is still small and experimental, but it already works for a restricted subset of pandas-like operations and runs deterministically in C++.
Repo: https://github.com/CVPaul/xpandas
I’d love feedback on whether this direction makes sense, and where people think it would break down.
Similar Projects
AI/ML●●●Banger
We built an LLM inference engine in pure Python – no PyTorch, no Triton
30x faster cold start than vLLM with zero PyTorch dependencies.
WizardryBig BrainZero to One
zyoraclub
202d ago
Developer Tools●●●Banger
Colnade – Type-Safe DataFrames for Python
Type-safe DataFrames without plugins: catch column typos in your editor, not production.
Big BrainSolve My ProblemWizardry
jwde
313mo ago
Developer Tools●●Solid
Nimic – write pure Python and compile AOT to native binaries via Nim
Python-to-Nim transpiler with ctypes-backed types when Cython and Numba already exist.
Big BrainNiche Gem
dima-quant
505d ago
Developer Tools●●Solid
Pyinc – From-scratch consistency in native Python
Frozen values and tracked reads prevent mutable state from breaking memoization.
Big BrainWizardry
brumbelow
111mo ago
Developer Tools●●Solid
Pyinc – a pull-based incremental query kernel for Python
Backdating equal results keeps downstream consumers valid without re-running.
Big BrainNiche Gem
brumbelow
101mo ago