diff --git a/.gitignore b/.gitignore
index 5a19a513..c02bb677 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,2 +1,3 @@
 .idea/*
-Manifest.toml
\ No newline at end of file
+Manifest.toml
+.minio*
diff --git a/src/AWSS3.jl b/src/AWSS3.jl
index 13dcb4db..624a39c7 100644
--- a/src/AWSS3.jl
+++ b/src/AWSS3.jl
@@ -18,6 +18,7 @@ export S3Path, s3_arn, s3_put, s3_get, s3_get_file, s3_exists, s3_delete, s3_cop
        s3_sign_url, s3_begin_multipart_upload, s3_upload_part,
        s3_complete_multipart_upload, s3_multipart_upload,
        s3_get_tags, s3_put_tags, s3_delete_tags
+export S3InputSerialization, S3OutputSerialization, s3_select_object_content, s3select
 
 using AWS
 using AWS.AWSServices: s3
@@ -839,6 +840,7 @@ function s3_nuke_bucket(aws::AbstractAWSConfig, bucket_name)
 end
 
 
+include("s3select.jl")
 include("s3path.jl")
 
 end #module AWSS3
diff --git a/src/s3path.jl b/src/s3path.jl
index 5fe860bf..405f14d1 100644
--- a/src/s3path.jl
+++ b/src/s3path.jl
@@ -382,3 +382,7 @@ function s3_get_name(prefix::String, s::String)
     tokenized = split(subkey, "/"; limit=2)
     return length(tokenized) == 2 ? first(tokenized) * "/" : first(tokenized)
 end
+
+function s3select(fp::S3Path, expression, args::AbstractDict{String,<:Any}=Dict{String,Any}(); kwargs...)
+    s3_select_object_content(fp.config, fp.bucket, fp.key, expression, args; kwargs...)
+end
diff --git a/src/s3select.jl b/src/s3select.jl
new file mode 100644
index 00000000..0755981b
--- /dev/null
+++ b/src/s3select.jl
@@ -0,0 +1,103 @@
+
+abstract type S3SerializationStruct end
+
+function _key_string(field::Symbol)
+    strs = split(string(field), '_')
+    join(uppercasefirst.(strs), "")
+end
+
+s3_serialization_dict(kwargs) = Dict(_key_string(k)=>v for (k,v) ∈ kwargs)
+
+Base.Dict(s::S3SerializationStruct) = s.dict
+
+
+"""
+    S3InputSerialization
+
+see AWS API docs: https://docs.aws.amazon.com/AmazonS3/latest/API/API_InputSerialization.html
+"""
+module S3InputSerialization
+    using AWSS3: S3SerializationStruct, s3_serialization_dict
+
+    """
+        S3InputSerialization.CSV
+
+    see AWS API docs: https://docs.aws.amazon.com/AmazonS3/latest/API/API_CSVInput.html
+    """
+    struct CSV <: S3SerializationStruct
+        dict::Dict{String,Any}
+    end
+    CSV(;kwargs...) = s3_serialization_dict(kwargs)
+
+
+    """
+        S3InputSerialization.JSON
+
+    see AWS API docs: https://docs.aws.amazon.com/AmazonS3/latest/API/API_JSONInput.html
+    """
+    struct JSON <: S3SerializationStruct
+        dict::Dict{String,Any}
+    end
+    JSON(;kwargs...) = JSON(s3_serialization_dict(kwargs))
+
+    """
+        S3InputSerialization.Parquet
+
+    see AWS API docs: https://docs.aws.amazon.com/AmazonS3/latest/API/API_ParquetInput.html
+    """
+    struct Parquet <: S3SerializationStruct
+        dict::Dict{String,Any}
+    end
+    Parquet(;kwargs...) = Parquet(s3_serialization_dict(kwargs))
+end
+
+"""
+    S3OutputSerialization
+
+see AWS API docs: https://docs.aws.amazon.com/AmazonS3/latest/API/API_OutputSerialization.html
+"""
+module S3OutputSerialization
+    using AWSS3: S3SerializationStruct, s3_serialization_dict
+
+    """
+        S3OutputSerialization.CSV
+
+    see AWS API docs: https://docs.aws.amazon.com/AmazonS3/latest/API/API_CSVOutput.html
+    """
+    struct CSV <: S3SerializationStruct
+        dict::Dict{String,Any}
+    end
+    CSV(;kwargs...) = CSV(s3_serialization_dict(kwargs))
+
+    """
+        S3OutputSerialization.JSON
+
+    see AWS API docs: https://docs.aws.amazon.com/AmazonS3/latest/API/API_JSONOutput.html
+    """
+    struct JSON <: S3SerializationStruct
+        dict::Dict{String,Any}
+    end
+    JSON(;kwargs...) = JSON(s3_serialization_dict(kwargs))
+end
+
+# for now this just looks at the key
+function _infer_input_serialization(aws::AbstractAWSConfig, bucket, key)
+    if endswith(key, "csv")
+        S3InputSerialization.CSV()
+    elseif endswith(key, "json")
+        S3InputSerialization.JSON()
+    else
+        S3InputSerialization.Parquet()
+    end
+end
+
+function s3_select_object_content(aws::AbstractAWSConfig, bucket, key, expression,
+                                  args::AbstractDict{String,<:Any}=Dict{String,Any}();
+                                  expression_type::AbstractString="SQL",
+                                  input_serialization::Union{Nothing,S3SerializationStruct}=nothing,
+                                  output_serialization::S3SerializationStruct=S3OutputSerialization.CSV())
+    input_serialization ≡ nothing && (input_serialization = _infer_input_serialization(aws, bucket, key))
+    S3.select_object_content(bucket, expression, expression_type, Dict(input_serialization), key,
+                             Dict(output_serialization), args, aws_config=aws)
+end
+s3_select_object_content(a...; b...) = s3_select_object_content(global_aws_config(), a...; b...)
diff --git a/test/select_experiments.jl b/test/select_experiments.jl
new file mode 100644
index 00000000..27832f0f
--- /dev/null
+++ b/test/select_experiments.jl
@@ -0,0 +1,8 @@
+using AWSS3, Minio
+
+s = Minio.Server(@__DIR__)
+
+csv = S3Path("s3://testbucket/test.csv", config=MinioConfig(s))
+
+test(csv) = s3select(csv, "select * from s3object s where s.C >= 4")
+
diff --git a/test/testbucket/test.csv b/test/testbucket/test.csv
new file mode 100644
index 00000000..92830226
--- /dev/null
+++ b/test/testbucket/test.csv
@@ -0,0 +1,6 @@
+A,B,C
+0.5231510582131029,kirk,1
+0.7377557522717386,spock,2
+0.1847573737369297,bones,3
+0.691671880638087,scotty,4
+0.30899820492077623,uhura,5